AI News Archive: August 28, 2026 — Part 4
Sourced from 500+ daily AI sources, scored by relevance.
- AI Has Human Doctors Asking: What’s Left for Us?
A recent paper argues that AI is often better at doctoring than doctors. Guess who isn't thrilled.
Score: 52🌐 MovesAug 28, 2026https://www.wired.com/story/ai-has-human-doctors-asking-whats-left-for-us/ - VA launches AI helpdesk chatbot, with plans for expansion later this year
The Department of Veterans Affairs is integrating more artificial intelligence to connect veterans with benefits. The post VA launches AI helpdesk chatbot, with plans for expansion later this year appeared first on FedScoop .
- The Week’s 10 Biggest Funding Rounds: AI Tools And Assistants Lead Sparser Lineup Of Megadeals
The Week’s 10 Biggest Funding Rounds: AI Tools And Assistants Lead Sparser Lineup Of Megadeals Crunchbase News
Score: 49💰 MoneyAug 28, 2026https://news.crunchbase.com/venture/biggest-funding-rounds-ai-tools-assistants-instinct/ - MAPS: Netflix’s Multimodal Asset Personalization at Scale
By Emma Yanyang Kong , Aditya Deshpande , Asad Abbasi , Bowei Yan , David Fagnan , Ashish Rastogi , Dhaval Patel , Ray Zhang Introduction The Netflix experience is a journey of discovery. Every visual cue, from the artwork on a title to the video previews that autoplay while you browse, is there to connect you with a story you will love. We call these visual cues assets , and choosing the right one for each member is a personalization problem of its own. But which image or video preview of Squid Game should we show you? And what do we do right after a title launches, when there’s far too little interaction data to know which asset we should recommend to each member? For years, our models answered the first question well and the second poorly. They learned which assets members interacted with, but treated every asset as an opaque ID, blind to what was actually in the artwork or video preview. Right after a title launched, its assets had no history, so we dialed up exploration on its assets to gather interaction data, and otherwise fell back to popularity heuristics that ignore your taste. Only once enough interactions had piled up could personalization take over. This is the classic cold-start problem. This post shares how multimodal embeddings let our models see and hear the assets they recommend, so personalization can kick in far sooner, close to a title’s launch. Because a new asset arrives with its embedding the model already understands, that embedding carries member taste signals from related assets immediately. Consequently, the model needs far less interaction history before it can personalize. We cover three production systems, artwork personalization, query-aware artwork ranking, and video preview personalization, plus a cheap trick for choosing new embeddings before committing to full end-to-end integration and A/B testing. Artwork Personalization A single image is often a member’s first touchpoint with a title, so we create a diverse set of artworks for each title to appeal to different member tastes. We already use personalized artwork based on members’ interaction histories, but this approach breaks down for newer titles and their assets, where there is little or no behavioral data to learn from. Making the Model See the Artwork Our solution is to let the model “look” at the picture. We encode each artwork with CLIP , a pretrained image-text embedding model, and fold the result into how the model represents that asset, concatenating the per-asset CLIP image embedding, a 768-dimensional vector, with the asset’s learned ID embedding to give an asset representation: e_id(a) is the asset’s learned ID embedding, and e_a is its CLIP image embedding. The two are concatenated and passed through an MLP layer to give h_a, the representation the model scores against a member. This single change transforms how the model handles a brand-new artwork. Instead of treating it as an unseen ID, the model now receives a CLIP embedding the moment the asset is created. That allows a member’s preferences over visual themes, talent, and color palettes to be applied immediately, long before the asset accumulates any interactions of its own. Because those preferences are expressed in image-embedding space rather than tied to specific asset IDs, they transfer seamlessly across titles . If you consistently engage with artwork featuring a particular comedian, the model can carry that signal to their new title and prioritize the asset that places them front and center, even if it has never shown you that exact image before, as in the figure below. In this way, cold-start shifts from being a blind spot to something the embedding space already has an informed opinion about. Knowledge transfer through CLIP embeddings. A member who has interacted with a comedian’s past stand-up artwork (left) leads the model to favor the new-title asset that features that comedian prominently (green check) over one that does not, even though it has never seen that specific image before. From Five Models to One That shift, from scoring an asset by the ID it happens to carry to scoring it by what the image actually contains, powers a second big win, model consolidation. Each title’s artwork spans multiple canvases with different croppings (billboard, vertical-box, horizontal-panel, short-panel, landscape-panel), and historically we trained a separate model per canvas, since an ID-based model has no way to know that the cropped and resized renderings of one scene are related, so signal could not flow between canvases and each faced its own cold-start. CLIP embeddings break that barrier. Because they are largely invariant to crop, resize, and aspect ratio, those near-identical renderings map to nearly the same vector, as the figure further below shows. A single unified model can therefore pool interaction signal across every canvas, so a member’s affinity learned on a high-traffic canvas immediately informs the artwork we pick on a sparse one. The result is one model in place of five, with the largest gains on the canvases that have the least interaction data. One source image, many canvases. The same Running Point artwork is cropped and resized across billboard, TV, mobile, and out-of-home placements, each with a different asset ID. Because CLIP embeddings barely change under crop and resize, a single unified model can personalize all of them. Mixing Five Canvases of Training Data Consolidation introduced a challenge that the per-canvas models never faced: how to effectively mix data across disparate canvases? The canvases differ widely in impression volume, and the interactions they log are not all worth the same to a member’s long-term experience. Training on pooled raw counts would let the highest-volume canvas and the most frequent interaction types dominate, so the low-data canvases we were trying to help would benefit least. Hand-tuning a weight per canvas would just trade that problem for a set of arbitrary hyperparameters and endless online sweeps to tune them. Instead we use reward-based weighting , building on Netflix’s long-term reward modeling . Each training example is weighted by the long-term reward score attached to its interaction type: a_ti is a training example, a positive interaction on asset i of title t. Its weight is set by the interaction type e observed on it, scored by ρ, that type’s long-term reward. where e (·) is the type of the observed positive interaction and ρ is that type’s long-term reward score. Because interaction types are not distributed evenly across canvases, weighting by long-term value rebalances the canvas mixture on its own, with no weight set by hand. A canvas contributes in proportion to the long-term value of the interactions it drives rather than to how many impressions it happens to get. Consolidation becomes feasible, and the unified model optimizes for long-term member satisfaction instead of whichever short-term action is most frequent. A Note on Offline Evaluation Every result presented here must clear two bars: an offline metric evaluation followed by a large-scale online A/B test. The offline metric is the subtle one. Judging a new model on logs from the current production policy is biased, because that policy shows some assets far more often than others. The logged rewards describe what the policy preferred, not what members would have chosen from the full candidate set, so a new model that disagrees with the logging policy looks worse than it is, because the impressions it would have picked are barely represented in the data. We handle this with inverse propensity scoring (IPS) computed on a dedicated slice of exploration traffic. A small fraction of traffic is served by a randomized policy that samples among a title’s candidate assets from a known distribution, so the propensity of showing a given asset in a given context is logged exactly at serving time rather than estimated after the fact. Reweighting every observation by the inverse of its logged propensity gives: where D is the exploration slice and r ( x , a ) is the observed reward, such as a play. Impressions that exploration made rare are upweighted accordingly, and the estimator becomes an unbiased estimate of the reward a candidate policy would have earned had we actually deployed it. Having propensities that are known by construction, rather than modeled after the fact, is in our experience the single biggest reason our offline numbers track online outcomes. We report IPS as a ratio against the production baseline, and a candidate has to win there before it gets any A/B traffic. Combining Both Ideas Works Better Two ideas are bundled together here, so we ablated them separately against the old five-model production system. V1, image embeddings only. The five per-canvas models kept as they were, each one augmented with image embeddings. V2, unified model only. A single model trained over all five canvases, but with learned ID embeddings alone and no image content. V3, both together. One unified model over all five canvases, with image embeddings in its asset representation. As the chart below shows, each idea helped exactly where we expected: on the data-starved short-panel canvas and landscape-panel canvas. V3 was the clear winner. A change inside ±1% is not significant for this offline metric, and those bars are hatched in the chart. Most of what V1 and V2 do on their own sits inside that band. Relative offline IPS lift by canvas for the three variants, each measured against the prior per-canvas model on that same canvas. Both ideas help where interaction data is scarcest, and V3 is strongest. Hatched bars fall inside the ±1% band, where the change in the offline metric is not significant; V3 values are labeled on the plot. In the online A/B test across all device platforms, which ran for at least four weeks, the results drew a much clearer line: Neither idea moved our online core member metrics on its own. V1 and V2 were both flat and non-significant, and only V3 won a statistically significant lift. It is what runs in production today. The two ingredients need each other. V1 tells a per-canvas model what an asset looks like, but one sparse canvas has too few examples to teach it how to use that. V2 supplies plenty of data, but only ID-based data, which a new asset lacks. V3 has both, so mature canvases teach the shared model how CLIP embeddings map to member preference and that mapping transfers straight to the sparse ones. The effects compound rather than add, since the V3 short-panel lift (5.691%) exceeds V1 and V2 combined. The lesson is to look for a second blocking factor before concluding that content features do not help. Cold-Start Challenge from a New UI Launch The real test came from the product change that motivated the work. Netflix was preparing its largest TV home-screen redesign in a decade , which would make short-panel the dominant artwork canvas effectively overnight. This was a cold-start problem in its sharpest form. The canvas about to receive the most impressions had the least historical data, and waiting for short-panel interactions to accumulate would have degraded the user experience. Consolidation lets short-panel selection draw on signal pooled from every canvas, and CLIP embeddings let the unified model personalize a short-panel asset that has gathered very few interactions of its own. We shipped V3 ahead of the launch and measured it with a month-long holdback A/B test, keeping a small control group on the prior per-canvas model. V3 absorbed the shift immediately, with statistically significant gains on both our core discovery metric and streaming hours, and larger gains than in the steady-state ablation. That stronger result is what we expected, since a sudden shift in which canvas dominates is exactly where V3 should help most. Query-Aware Artwork Personalization Your general taste is the right signal when browsing, but not when searching . For example, when searching for a specific actor, you want artwork that features them, even if your broader taste says otherwise. On the Netflix Search Page, the member’s intent is explicit and stated in the query, and the displayed artwork should reflect it. The same CLIP embeddings we added for cold-start hand us this almost for free. Because CLIP projects text and images into one shared embedding space, we can measure how well a query matches a candidate artwork directly by the cosine similarity between the CLIP text embedding of the query and the CLIP image embedding of the asset. We blend that alignment term with the usual personalization score: Here the personalization term is the score the artwork model above already produces for a member and asset, the second term compares the text embedding of the query against the image embedding of the asset, and the mixing weight α between 0 and 1 is tuned through online A/B testing. The first term is “what we think you like”; the second is “what you just asked for,” and α sets how much each matters. Crucially, this took no extra modeling effort. The CLIP embeddings already sit in the asset representation from the artwork work above, so they carry the text-image alignment for free, and we get a query-aware ranker by adding a single similarity term at scoring time. The effect is visible in the search results themselves. Query-aware artwork for a search for a specific actor. Each result surfaces an asset that visually features the searched actor, aligning the artwork with the member’s explicit intent. Personalizing Video Previews via MediaFM Video previews raise the bar over still artwork. A video preview unfolds over time, and its appeal comes as much from motion, pacing, dialogue, and soundtrack as from any single frame. Our older video preview personalization models saw none of that. Like the early artwork models, they treated each preview as an opaque ID. Our first content-aware attempt, SeqCLIP , described a video preview by its frames, encoding each with a CLIP embedding and then averaging them into one vector. That captured what a video preview looked like, but a mean of still frames still misses what it sounds like, the dialogue and music that carry so much of a preview’s tone. To capture the rest, we turned to MediaFM , Netflix’s first in-house multimodal foundation model. Trained on 80 million shots, MediaFM fuses the following three signals per shot into a single embedding: Visual: SeqCLIP Audio: A pretrained speech and audio embedding model Text: Captions encoded via a large-scale text model Adopting MediaFM required no new infrastructure, since we simply integrate its shot embeddings into the asset representation, exactly as we did with CLIP embeddings for artwork. The added modalities paid off. We evaluated both embeddings against the ID-only baseline offline with IPS and then in a five-week online A/B test across all device platforms, and both signals gave the same ordering, MediaFM > SeqCLIP > ID-only, and each step of added content awareness helped, with the gains largest on TV. Offline, both content-aware embeddings beat the ID-only baseline on IPS and MediaFM beat SeqCLIP, as the chart below shows. Online, MediaFM came out on top too, delivering a statistically significant lift in our core streaming metric over the ID-only baseline and outperforming SeqCLIP. This shows that the audio and timed-text signals, which a visual-only encoder like SeqCLIP cannot capture, add real value. We have since shipped MediaFM as the default video preview embedding across all platforms. Relative offline IPS lift for the two content-aware video preview embeddings, each measured against the ID-only baseline at the zero rule. Adding visual content awareness helps, and adding audio and timed text on top of it helps further. Choosing Embeddings Cheaply with a Proxy Task New embeddings arrive constantly, but end-to-end trials are expensive, which cost data engineering, model retraining, and weeks of A/B test traffic. We couldn’t afford to run the full pipeline for every candidate, so we gated the funnel with a cheap question: From the content embedding alone, can you predict which asset wins under a plain, unpersonalized policy? We first select a fixed set of titles. For each title we use exploration data to find its debiased popularity winner, the asset with the highest interaction rate after we adjust for how often it was shown using its propensity score. We mark this winner with a binary label, 1 for the winner and 0 otherwise. We then train a linear probe to recover that label from the asset embedding alone, with no title, cast, or metadata, by minimizing the standard binary cross-entropy loss: Keeping the probe linear and embedding-only is intentional, since it isolates how much of an asset’s popularity is actually encoded in the embedding. If the embedding captures the semantic drivers of popularity, a simple linear classifier should be able to identify likely winners. If it does not, the probe performs no better than random guessing, which is the baseline we score it against. We first used the linear probe to screen and prune a broad set of candidate embeddings before modifying any production pipeline, narrowing the field to two finalists, SeqCLIP and the leading MediaFM variant. We then carried both through full offline evaluation and online A/B testing. All three signals, the linear probe accuracies, the offline IPS lifts, and the online A/B results, ranked MediaFM ahead of SeqCLIP, as the chart below shows. That alignment is why the linear probe now gates every new MediaFM version before release. Linear probe Δaccuracy, offline IPS lift, and online A/B metric lift for the two finalists. All three agree that MediaFM beats SeqCLIP. The online panel is measured against the ID-based baseline, with its values withheld. The Netflix Embedding Store None of this would be practical without shared infrastructure. Every embedding in this post, CLIP for artwork, SeqCLIP and MediaFM for video previews, lives in the Netflix Embedding Store , a component of Netflix’s AI Platform that hosts dense embeddings for titles, games, member profiles and multimedia assets. A foundation model encodes raw asset content into a dense vector once, and the Embedding Store serves that vector to every downstream system, the artwork model, the query-aware ranker, the video preview model, and others, through the same interface. Crucially, it serves the exact same embeddings at training time and at online inference time, so there is no skew between what a model learns from and what it sees in production. Its key property is that it decouples foundation-model updates from personalization-model deployments . A new embedding, or a new version of an existing one, can be registered, backfilled across the catalog, and validated entirely on its own, without touching the training or serving code of any model that consumes it. Once it is in the Embedding Store, it becomes available to every ranking and personalization model through configuration alone, no downstream code changes, no coordinated release. This is what let us swap CLIP into the artwork model, stand up the query-aware ranker on the same vectors, and roll MediaFM through the video preview model, each as an independent change rather than a cross-team migration. Foundation-model embeddings (CLIP, SeqCLIP, MediaFM) are stored once and consumed by every downstream system: artwork, query-aware artwork, video previews, and other rankers. What We Learned, and What’s Next Three lessons stood out. Pretrained CLIP embeddings let us consolidate five artwork models into one while boosting performance on data-starved canvases. This benefit became especially clear when the redesigned TV home screen rolled out. For video, multimodality wins decisively. The audio and text signals that a purely visual encoder cannot access pushed MediaFM past SeqCLIP. A cheap proxy task yields big savings , efficiently pruning the candidate set before running full end-to-end experiments and online A/B tests. Next, we aim to extend the Embedding Store toward a single shared semantic space for image, text, and video . Such a unified representation would enable cross-modal retrieval, such as matching a video preview to a search query, or a static artwork to the video preview it was derived from, as well as unified asset ranking across surface types and a more cohesive, intuitive discovery experience for members everywhere. Acknowledgements We thank Aneesh Vartakavi , Santiago Castro , and Avneesh Saluja for the CLIP embedding and MediaFM work that made the content-aware models described here possible, and Ratna Kavuri for the backend systems that serve multimedia personalization in production. MAPS: Netflix’s Multimodal Asset Personalization at Scale was originally published in Netflix TechBlog on Medium, where people are continuing the conversation by highlighting and responding to this story.
- Hanoi approves 24-month trial of autonomous buses, robotaxis and delivery robots
Hanoi has allowed Phenikaa-X JSC to conduct controlled testing of a smart autonomous transport system, including autonomous buses, robotaxis, automated cleaning robots and delivery robots, at Hoa Lac Hi-Tech Park and the Hoa Lac campus of Vietnam National University, Hanoi.
- Brain-reading earbuds are already here, prompting urgent questions about children and neurodata
While it may sound like science fiction, earbuds that read brain activity are already on sale. Wearable technology company Neurable claims that its headphones, equipped with electroencephalography (EEG) sensors, can track your fatigue and reduce burnout. At night, NextSense's Smartbuds pick up the electrical signals of your brain to chart your sleep. By day, the same EEG sensors help schedule demanding tasks for the hours when you are sharpest.
Score: 49🌐 MovesAug 28, 2026https://techxplore.com/news/2026-08-brain-earbuds-prompting-urgent-children.html - ARIA Bans Fully AI-Generated Songs From Australian Music Charts
ARIA now requires Australian chart-eligible music to be substantially human-made, excluding songs where AI creates the primary creative elements. The post ARIA Bans Fully AI-Generated Songs From Australian Music Charts appeared first on TechRepublic .
Score: 49🌐 MovesAug 28, 2026https://www.techrepublic.com/article/news-aria-ai-generated-music-australia-charts-apac/ - Warnock blasts OpenAI, Effingham for doing data center deal ‘in the dark’
Warnock blasts OpenAI, Effingham for doing data center deal ‘in the dark’ AJC.com
- How AI is helping keep critical medicines on pharmacy shelves
How AI is helping keep critical medicines on pharmacy shelves USA Today
- Renesas opens Beijing physical AI, robotics lab
The facility will let customers demonstrate, validate, and jointly develop next-generation robotic systems as Renesas expands its physical AI efforts.
Score: 49🌐 MovesAug 28, 2026https://www.techinasia.com/japan-tech-earnings-jump-on-ai-demand-in-data-centers - Honda, Nissan to reach deal on joint development of vehicle software
Honda, Nissan to reach deal on joint development of vehicle software Nikkei Asia
Score: 49🌐 MovesAug 28, 2026https://asia.nikkei.com/business/automobiles/honda-nissan-to-reach-deal-on-joint-development-of-vehicle-software - The data center boom will fail its promise if safety and workforce development continue to lag behind
The data center boom will be judged not only by how much infrastructure we build, but by whether the people building it return home safely. Major opportunities bring a rush for capital, market share, and industry leadership. Schedules tighten, hiring accelerates, and safety can be treated as an obstacle. From what I have seen in […] This story continues at The Next Web
Score: 49🌐 MovesAug 28, 2026https://thenextweb.com/news/data-center-boom-safety-workforce-development - Shadow agents make the old shadow IT problem worse
Companies are running into a new version of an old problem. Take, for example, a marketing executive turns on an agent from Marketo but doesn’t tell their IT team. That shadow agent now has access to customer data with little oversight and no clear adherence to the company’s security or compliance policies. That’s a real scenario that’s happening now. It’s also the same pattern that made shadow IT a headache for over two decades. Employees often adopted unapproved tools that helped them do their job faster and better because the company option just wasn’t as good. With agentic AI, it’s different because of what an unauthorized agent can do. Shadow IT tools mostly accessed or handled data without proper authorization. Agents, on the other hand, can take it much further by acting on that data by pulling records, sending messages and making changes with little oversight. Forcing agent data interactions via model context protocol can help, but the fact is there’s an autonomous agent that IT has little to no visibility into. Now multiply that one marketer by every department, plus every agent that your vendors in HR, finance and elsewhere run in your systems. It’s a situation that most companies have no way to see, let alone control. This can become hundreds of vendors each running multiple agents, moving between your company, their company and your supply chain. That’s many thousands of agents with no consistent way to track what they are doing. I often ask security teams: Do you have a robust inventory of the agents operating on your network? For most companies, the answer is no. And even if a company scanned its network, it’s not clear that agents are what they claim to be. So, there can be an unknown number of agents from unverified provenance performing a multitude of tasks and data exchanges on your network. This unfortunately is the current state of affairs at most companies. Old problem, higher stakes Shadow IT used to be technology that was used for work without explicit IT approval. Ten years ago, that meant a personal Dropbox folder or an unsanctioned project board. Today, it can mean an agent with a login and a task list, who is working inside your systems. That’s what shadow agents are. When an agent gets access, it can quickly pull a record, draft a reply, update a field, move a file — the list goes on. This can happen before anyone even notices anything happening. Security teams often say you can’t protect what you can’t see. That was true when the invisible thing was a spreadsheet. But it reaches another level when that agent has a login and knows what to do with it. The precedent: DMARC solved this for email How do you know an email that looks like it’s from Uber is actually from Uber? I ran into this problem, years before AI was an issue. Say you get an email after an Uber ride saying, “Thanks for riding, click here for your receipt.” It says it’s from Uber. It’s actually sent by a vendor like SparkPost, on Uber’s behalf. Uber authorized it. The vendor is doing what Uber asked it to do. But nothing in the email told your inbox that there was permission, so your inbox just trusted the “from” line. Worse, it could’ve easily been a phishing attack from a criminal purporting to be Uber. To address this, the email sender should have been authenticated against a trusted source before the email lands in your inbox. That’s DMARC. A decade and over 1 million domains later, it’s proof the model works at scale. Verify the sender against a record that the domain owner controls and the guessing goes away. The same fix, one layer up, for agents DNS is already the internet’s phone book and a secure, trusted and public-facing database representing the domain. An agent claiming to represent Salesforce or any vendor can be checked against Salesforce’s DNS-secured record before it’s ever granted access. That one check answers a critical question: Is this agent who it says it is, and did the company it claims to represent actually authorize it? Once an agent is verified via the domain owner’s DNS record, there’s a domain cryptographically attached to the agent, and accountability that didn’t exist before. Right now, most organizations don’t have that. An agent shows up and asks for access. But if something goes wrong, no one is really responsible, because nobody checked in the first place. They “trust” that the agent is what it claims to be. This gets even more complex if the agent is from a known hyperscaler/AI company but acting on behalf of an untrusted/unknown user (e.g., a ChatGPT agent but getting instructions from a criminal) Zero trust as a bouncer, not a detective Most of security has worked the same way for years: Try to identify everyone who shows up, then decide if they’re trustworthy. That’s backwards, and email proved it over a decade ago. A layered approach with zero trust upfront and further interrogation of what’s left over combines the efficiency and low cost of zero trust with deep inspection as a second pass, providing a highly effective level of security. Layer 1: Zero trust is like a bouncer at a nightclub. She doesn’t try to figure out which of the world’s 8.3 billion people you are. She checks a finite short list of a couple dozen guests. If you’re not on it, you don’t get in. Layer 2+: Now that we’ve reduced the number of people by (usually) orders of magnitude, we can, if needed, run the remaining people through deeper inspection, say a metal detector, access credentials and so on. With agents, the logic is the same: Don’t evaluate whether an agent is trustworthy after it’s already inside your systems. Check whether it’s on the list before it gets anywhere near the door. This means a legitimate agent might get turned away because someone forgot to add it to the list. But that inconvenience is much better than the alternative of letting everything in and just hoping things go well. Identity is not permission There are two separate questions inside every access decision. Most conversations about AI governance conflate them. First, who is this? Second, what are they be allowed to do? This is like a passport and a visa. The passport says who you are. The visa says what you’re permitted to do and where you’re permitted to go. They’re issued by different authorities for different reasons. Confusing the two is where a lot of security approaches go wrong. Upfront agent authentication, also known as a passport, can tell you whether that agent claiming to be from Salesforce is really Salesforce’s. Next, you need to figure out if Salesforce is authorized to work in your system and what it’s allowed to do. Your approved list of vendors and the approved actions should be made on purpose rather than defaulting to whatever the agent claims about itself. The identity layer has to get solved first and solved the same way for everyone. But the permission layer is where every company’s answer is different, based on what that specific agent actually needs to access. Complicating matters, agents can change their workload mid-process. So, the permissions need to be continuous and focus on ongoing workloads as they evolve. This is a growing, urgent problem. Why this matters now AI-generated phishing is now three times more effective than traditional campaigns, according to Microsoft . A year ago, AI-written phishing attempts were easy to spot due to bad grammar or strange tone. That has changed on an exponential curve, a progression humans’ brains have a hard time grasping. The same category of tool getting better at impersonation is now showing up as unauthorized agents inside company systems. And improving exponentially. Better deception plus more access adds up to a problem that grows faster than we can even imagine. Shadow IT taught the industry a lesson. Now, shadow agents are teaching it again. You can’t secure what you don’t know is running inside your systems. For the marketer turning on the Marketo agent, what would have caught it isn’t a smarter firewall or a longer policy document; it’s a check, run before access is granted, confirming that the agent is who it claims to be and that someone actually authorized it. It’s followed up with continuous permissioning and logging to make sure the agent does what it’s supposed to. That’s the shift: Verify an agent’s identity before it gets anywhere near the door. Email already proved this model at scale across roughly 1 million domains. Shadow agents are the same problem showing up again in a new form. It doesn’t need a new fix. It needs the one that already works.
Score: 49🌐 MovesAug 28, 2026https://www.cio.com/article/4214992/shadow-agents-make-the-old-shadow-it-problem-worse.html - Gemini Notebook Can Now Dig Into eBooks You Bought on Google Play
Gemini Notebook Can Now Dig Into eBooks You Bought on Google Play PCMag
Score: 49🌐 MovesAug 28, 2026https://www.pcmag.com/news/gemini-notebook-can-now-dig-into-ebooks-you-bought-on-google-play - Vietnam pushes Qualcomm, Samsung Electronics for bigger AI, chip investments
Vietnam has urged Qualcomm and Samsung to expand investments in AI, semiconductors and R&D as it seeks to position itself as a regional innovation hub.
- AI's Game Changing Impact on Financial Services E
AI's Game Changing Impact on Financial Services E IT Pro
Score: 48🌐 MovesAug 28, 2026https://www.itpro.com/technology/artificial-intelligence/ais-game-changing-impact-on-financial-services-e - The real reason AI isn’t paying off in finance
If you work in finance, you’ve probably been handed an AI tool in the last year or so. Maybe a copilot in your spreadsheet, maybe something bolted onto the close, maybe a chatbot that promised to answer any question about the numbers. And maybe, if you’re being honest, it hasn’t changed your Tuesday very much. You’re not doing it wrong. The tool isn’t broken. What’s missing is the part nobody put on the slide: AI is only as good as the work it’s standing on, and most of the time, the work underneath it is a mess. Confident AI answers you can’t trust Here’s a familiar scene. Someone asks the AI assistant a reasonable question — “why did margin move in the East region last month?” — and it produces an answer that sounds great. Confident. Well-organized. Possibly even formatted with little bullet points. The only problem is that you have no idea whether it’s right, because you don’t know which data it pulled, whether it used the current cost allocation method, or whether it quietly grabbed last fiscal year’s calendar. So you do what any sensible finance person does. You check it by hand. Which means the AI didn’t save you the work. It added a step. This is the quiet truth about why so much finance AI stalls. It’s not that the models can’t reason. It’s that they’re reasoning over data that was never cleaned, rules that were never written down, and logic that lives in one analyst’s head and three tabs of a workbook nobody else can open. AI didn’t create that gap. It just made it impossible to ignore, because now something is making decisions on top of it. McKinsey looked at how finance teams are actually using gen AI and found a useful counterexample. Across the handful of finance functions where they saw AI adopted in earnest, professionals were spending 20 to 30% less time crunching data — and putting that time back into the analysis their job is supposed to be about. In one case, a global consumer goods company pointed a gen AI assistant at budget-variance work and saw roughly 30% of that manual effort disappear. That’s a real result. But notice what made it real: it was pointed at a specific, repeatable task, working from data the team had already organized around a shared definition of what “variance” even means. The AI didn’t figure that out on its own. The team handed it a problem that was ready to be automated. What separates the workflows that pay off The finance work where AI delivers tends to share a few traits. It’s bounded — a clear start and end, not “answer anything about the business.” It’s repeatable, the same shape every month. And it’s tied to something that matters: cash, margin, risk, a number someone downstream is going to act on. That’s the easy part to say. The harder part is what has to be true underneath. For AI to work on one of those tasks, the data feeding it has to be prepared and validated before the model ever sees it. The rules — what counts, what gets excluded, how things roll up — have to be defined by your team and applied consistently, not guessed at by a model that’s never read your policy manual. And when the output lands, you have to be able to trace it back: which numbers, which logic, who signed off. In finance, that traceability isn’t a nice-to-have. It’s the difference between an answer you can put in front of an auditor and one you can only put in front of people who won’t ask hard questions. Think about the difference between two versions of the same workflow. In one, the AI reaches into raw data, applies whatever it infers the rules to be, and gives you a number. In the other, the data gets cleaned and structured first, your team’s actual business logic gets applied to it, and only then does AI work on top of a foundation it can stand on. The first one feels faster right up until something’s wrong and you can’t tell why. The second one is the one you can defend in a meeting. That’s really the test worth applying to any AI effort on your desk: can you explain where the answer came from, and would the explanation survive someone pulling on it? If yes, you’ve got something worth scaling. If no, more AI won’t fix it — it’ll just produce wrong answers more quickly. Where this leaves you on Monday None of this means starting over. The business logic your team has built — the spreadsheets, the rules, the institutional memory of how things actually work here — is the valuable part. The goal isn’t to throw it out for an AI that doesn’t know any of it. It’s to get that logic into a form that’s governed and repeatable, so AI can finally do something useful with it. The most practical move is also the least dramatic. Pick one workflow. Not the whole close, not “AI across finance.” One bounded, repeatable, annoying task you’d happily never do by hand again — invoice matching, a recurring variance pull, a report you rebuild every month. Get the data right for that one thing, write the rules down, and put AI to work on top of it. When it works, you’ll have something real: a workflow you can trust, and a clear sense of what the second one should be. Two traps worth naming, because they’re the ones McKinsey watched teams fall into. One is waiting for perfect data before you do anything — you’ll be waiting forever, and the team next door will have shipped three workflows by the time your data is pristine. The other is the opposite mistake: automating a process that’s still a tangle of exceptions and one-offs. Drop AI on top of a fragmented workflow and it doesn’t simplify it, it just adds a confident-sounding layer to the mess. The move is in between: standardize the one thing first, then automate it. If you want a low-stakes way to see what that looks like before you commit, our AI-Ready Starter Kits are built for exactly this. AI-Ready Starter Kits are pre-built Alteryx workflows and synthetic datasets designed to demonstrate how Alteryx can be applied to specific business use cases. They prepare and structure data to produce analysis-ready outputs, which can be extended using external AI tools such as large language models (LLMs). They won’t run your finance function — that’s not what they’re for. But they make the shape of a workflow that actually pays off tangible enough to copy. The AI on your desk isn’t the problem. The work underneath it is. Fix that for one thing, and you’ll stop wondering why AI hasn’t paid off — because it finally will. Search our full AI-Ready Starter Kit library to find finance use cases fit for you and your team. To learn more, visit us here .
Score: 48🌐 MovesAug 28, 2026https://www.cio.com/article/4212781/the-real-reason-ai-isnt-paying-off-in-finance.html - AI’s Game-Changing Impact on Manufacturing
AI’s Game-Changing Impact on Manufacturing IT Pro
Score: 48🌐 MovesAug 28, 2026https://www.itpro.com/technology/artificial-intelligence/ais-game-changing-impact-on-manufacturing - UW AI researcher wins top Marconi award and aims to put ‘superhuman hearing’ in billions of devices
Ph.D. student Malek Itani, who co-founded the AI-powered sound enhancement startup Hearvana, was recognized for developing on-device AI algorithms that enable real-time target speech extraction, "sound bubbles," and semantic hearing on low-power hearables like earbuds and hearing aids. Read More
- How internal deployments build trust in enterprise technology
For marketing leaders, trust is not built by messaging alone. Every enterprise technology investment asks customers to rely on new technology for critical business processes. Before they make that kind of commitment, they need confidence in both the technology and the company behind it. That makes credibility especially important. Product messaging can explain what technology is designed to do, and customer examples can show results in other organizations. But customers can tell when a company truly believes in what it sells. There is a different level of confidence that comes from knowing the people behind the product depend on it themselves. At Rocket, using our own products to run parts of our business turns confidence into visible proof. Our technology operates in real workflows and faces many of the same demands our customers encounter. That experience strengthens the connection between what we promise in the market and how we show up as a company. Experience makes brand promises more credible Leaders evaluating enterprise technology often hear similar claims from competing vendors. Enterprise technology always comes with ambitious promises: greater efficiency, stronger visibility, and better business outcomes. What leaders really want to know is whether the technology has proven itself where the work actually happens. A company that uses its own products can speak from direct experience. Teams working with customers develop a clearer understanding of how the technology fits into everyday operations, making those conversations more practical because they are grounded in lived use, not theory. Sales and marketing teams can explain the technology’s value through real examples and outcomes. Service and product teams can respond with greater empathy because they understand, firsthand, what it means to depend on the technology in the flow of daily work. Internal adoption in practice Following Rocket’s acquisition of AMC, our legal team inherited thousands of contracts that required review. The volume placed significant pressure on a process that already depended heavily on manual work. The team initially considered purchasing a contract lifecycle management platform. Then another question emerged: Could technology already developed by Rocket solve the problem? Rocket’s legal team worked with the company’s Mobius product team to adapt Rocket Mobius and Smart Chat for the contract-review process. The deployment transformed a large collection of contracts into searchable information, while AI-assisted metadata extraction made important details easier to locate. For Rocket’s legal team, the immediate impact was immediate and measurable. The deployment: Reduced initial contract review time by 80% Shortened acquisition-related contract reviews from weeks to about an hour Centralized contracts in a searchable repository with AI-assisted metadata extraction and indexing Reduced reliance on outside counsel, allowing the legal team to focus on higher-value legal work. Just as importantly, the solution proved repeatable. What began as a response to the AMC acquisition evolved into a scalable process for acquisition-related contract reviews and has been used in every acquisition since enabling the legal team to assess large contract portfolios quickly and consistently. The results mattered, but they were not the only takeaway. The Mobius solution supported a business-critical process where performance directly affected employees’ ability to complete important work. That choice creates a different kind of evidence. The technology addressed a genuine enterprise need inside the company that developed it, under conditions that could not be simplified or controlled for a demonstration. Internal adoption also gives employees firsthand experience with the same technology customers depend on. That shared experience creates a stronger connection between the people who build and support the product and the customers who use it every day. Internal use strengthens the brand Trust grows when what a company says matches what it does. Marketing can establish the promise, and product performance can substantiate it. Internal adoption shows that the organization believes in that promise strongly enough to depend on the technology itself. For a brand, that alignment is powerful. People notice when a company’s market message is reflected in its own behavior. Internal use strengthens customer assurance by creating greater consistency across customer interactions. When teams across the business share firsthand experience with the technology, customers receive clearer, more credible guidance about what adoption actually involves. It also creates a stronger sense of accountability. When employees inside the company depend on the same technology, the vendor shares more directly in the consequences of the experience. Reliability and continued value become internal priorities as well as customer commitments. Becoming your own customer In markets where competing technology providers make similar claims, credibility becomes an important source of differentiation. Enterprise leaders may see comparable features across multiple vendors, but the confidence they place in those vendors can vary significantly. Internal adoption gives customers another reason to believe. It shows that the company understands the operational consequences of the decisions its customers are making because it has put its own technology to work where business outcomes matter. That is where internal adoption becomes more than an operational decision. It becomes part of the brand itself. That is the value of becoming your own customer. Learn more here .
Score: 48🌐 MovesAug 28, 2026https://www.cio.com/article/4215400/how-internal-deployments-build-trust-in-enterprise-technology.html - Fast, fault-tolerant PyTorch training on AI Runtime
At scale, your training efficiency is determined by a single metric: "goodput", the...
Score: 48🌐 MovesAug 28, 2026https://www.databricks.com/blog/fast-fault-tolerant-pytorch-training-ai-runtime - Prompt: The AI Infrastructure Boom Is Getting Bigger Than GPUs
Nvidia’s record quarter shows AI demand is still surging as the infrastructure race expands into CPUs, networking, robotics and edge computing.
Score: 48🌐 MovesAug 28, 2026https://aibusiness.com/generative-ai/prompt-ai-infrastructure-boom-getting-bigger-than-gpus - Chart of the Day: China’s AI Chipmakers Are Turning Self-Reliance Into Sales
Chart of the Day: China’s AI Chipmakers Are Turning Self-Reliance Into Sales Caixin Global
- Waymo in Munich: Could autonomous vehicles make Europe's roads safer?
Waymo is bringing driverless taxis to Munich, its third city outside the US, as data suggests autonomous vehicles are far safer than human drivers.
Score: 48🌐 MovesAug 28, 2026http://www.euronews.com/next/2026/08/28/waymo-in-munich-could-autonomous-vehicles-make-europes-roads-safer - Enhancing reproducibility in hybrid Earth system models
Nature Machine Intelligence, Published online: 28 August 2026; doi:10.1038/s42256-026-01299-5 AI integration in Earth system models enhances prediction and modelling capabilities but also amplifies challenges for reproducibility. This Perspective introduces a framework for assessing reproducibility and provides practical ways to strengthen reproducibility in hybrid Earth system models.
- How AI Data Centers Are Making Everything More Expensive
How AI Data Centers Are Making Everything More Expensive Business Insider
Score: 48🌐 MovesAug 28, 2026https://www.businessinsider.com/how-ai-data-centers-are-making-everything-more-expensive-2026-8 - AI’s Game-Changing Impact on Retail E
AI’s Game-Changing Impact on Retail E IT Pro
Score: 48🌐 MovesAug 28, 2026https://www.itpro.com/technology/artificial-intelligence/ais-game-changing-impact-on-retail-e - Hyundai Mobis shares AI manufacturing technologies with suppliers
Hyundai Mobis shares AI manufacturing technologies with suppliers 매일경제
- Is the frontier of humanoid robotics shifting from hardware to intelligence?
As robots move into real-world deployment, intelligence is becoming as important as physical capability, Galbot’s founder says.
Score: 48🌐 MovesAug 28, 2026https://kr-asia.com/is-the-frontier-of-humanoid-robotics-shifting-from-hardware-to-intelligence - New report warns AI data center boom squeezes memory-chip supply, could raise everyday tech costs
The AI data center boom could make everyday technology more expensive, a new policy report warns, as chipmakers shift scarce memory-chip capacity toward higher-margin AI uses.
- Reinventing software engineering for the AI-first enterprise
AI is reshaping software engineering from task-level automation into an intent-driven, AI-orchestrated lifecycle spanning design, development, testing, security, deployment, and maintenance. To unlock its full value, enterprises must redesign delivery models, governance, platforms, and teams combining AI’s speed and scale with human judgment, domain expertise, accountability, and trust. This shift will also redefine engineering roles, with developers and managers increasingly becoming orchestrators of AI-enabled outcomes rather than executors of individual tasks. The leaders will be those that move beyond using AI as a productivity tool and make AI-first engineering an operating foundation for continuous, outcome-led enterprise transformation.
- The logic layer: the missing piece in modern AI tech stacks
There’s a scenario that plays out every day across the enterprise. A salesperson is about to close a major deal. They want to know what their commission will be. They type the question into ChatGPT or their favorite AI assistant. What comes back is a thoughtful, well-written explanation of how software companies typically structure sales compensation. The one thing it won’t tell them is what their commission will actually be if they close this specific deal. That gap between what AI can reason over and what it knows about your business is the defining challenge of enterprise AI adoption right now. I call it the logic layer. And without it, AI gives you impressive sounding outputs that are often disconnected from how your business runs. Why business logic lives with the analyst One of the more persistent myths in AI is that analysts are on the verge of becoming unnecessary. The reality is the opposite, and the logic layer is exactly why. In an AI-enabled enterprise, analysts become more essential because they are closest to the logic and context that governs the business. They know which definition of pipeline matters and which edge cases matter in audit, merchandising, finance, or marketing. I believe enterprises that succeed in the AI era will not be defined by how much AI they deploy but whether the people who understand the business own and control the intelligence that runs it. If that ownership defaults entirely to IT or to a vendor’s black box, companies risk scaling systems they cannot fully adapt or audit. Giving business teams the tools and mandate to own their logic is what makes the AI system trustworthy and responsive to how your business runs. That is why I see analysts as the architects of this next phase. What the logic layer looks like in practice Let me return to the commissions example, because it illustrates the concept precisely. Right now, when a salesperson needs to know their commission on a deal, they send a message to the commissions analyst. That analyst has their own spreadsheet — because comp plans change every quarter, with spiffs and special programs layered on top. They run the math manually and send back an answer. What if that same analyst built a simple, well-defined calculator that encoded their commission logic — the actual rules for your company, your plans, your programs — and connected it to the AI systems your salespeople are already using? Now when a rep asks what their commission will be on a specific deal, they get the right answer. Not a generic explanation of how commissions work. And here’s the compounding value: that same logic can then be used by the annual planning agent to model the operational cost implications of different comp plans. It can feed the scenario planning model that runs hundreds of simulations for financial planning. The analyst who built it enables an entire network of AI systems to act on accurate, business-specific logic. That’s the logic layer in practice: curated, purpose-built data assets and calculators that encapsulate how your business works, maintained by the people who understand it, deployable to every AI system that needs it. What the logic layer requires This is where I think most companies are still stuck. They’ve made the infrastructure investments. They have cloud data platforms and approved LLMs. But they’re asking those systems to do things they were never designed to do on their own. The logic layer requires three things: Purpose-built data assets. A narrow, clean, well-defined data set that reflects how you actually measure a specific business process. Encoded business logic. This is the part that lives in people’s heads right now — the policies, the edge cases, the context that makes data mean something. The ability to update it. Nobody runs a business to keep it the same. The logic layer has to be something that domain experts can update when the business changes. A pragmatic path forward The good news is that you don’t have to wait for a perfect architecture before you start building a logic layer. Start with your highest-value, most-repeated business processes — the ones where an analyst is currently fielding the same questions week after week. These are the processes where encoding logic into a curated, AI-ready data asset delivers immediate, measurable value. Then, empower your analysts to own that encoding — not IT. Give them low-code tools to do the work, and the mandate to treat that encoded logic as a strategic asset they own and evolve as the business changes. This is also where leadership posture matters. I have said for a while that this should not be framed as a choice between business and IT. It is both. IT should set standards, manage infrastructure, establish security boundaries, and make approved AI capabilities available across the organization. But IT should not become the bottleneck for every piece of business logic the company needs to operationalize. If this feels familiar, it should. We have seen this pattern before in enterprise technology. Infrastructure and platforms matter. But the last mile, the part that turns capability into business value, always depends on the people closest to the work. AI is no different. The companies that get the most from AI will be the ones that treat it like an operating model. They will automate core workflows, curate the right data, and empower analysts and domain experts to define the logic that makes AI useful and generate answers the business can use. I recently had a chance to go deeper on these ideas on the Talking AI podcast . If you want to hear more of my thinking on the analyst’s evolving role, how the logic layer connects to agentic workflows, and why I think the next 18 months will be pivotal for getting this right, it’s worth a listen. To learn more, visit us here .
Score: 47🌐 MovesAug 28, 2026https://www.cio.com/article/4215189/the-logic-layer-the-missing-piece-in-modern-ai-tech-stacks.html - Deploy an Open Model from Checkpoint to Inference in Two Commands with NVIDIA TensorRT Model Connect
Open AI models are evolving faster than ever, but bringing them into native applications can still require model-specific conversion, preprocessing,...
- Dell’s NVIDIA-based AI powerhouses compared: The Dell Pro Max with GB10 versus the Dell Pro Max with GB300
Dell’s NVIDIA-based AI powerhouses compared: The Dell Pro Max with GB10 versus the Dell Pro Max with GB300 IT Pro
- Maastricht-based Clementine raises €1.7 million to scale its AI-powered hearing care platform
Clementine, a Maastricht-based startup that has built an AI-powered platform to help hearing clinics convert leads into patients, has raised €1.7 million ($2 million) in Seed funding. The Seed round was led by Healthy.Capital, an Amsterdam-based early-stage HealthTech investor that has backed Pharmi, MS Sherpa, OpenUp, Datamonk, Clear.bio, among others. “Clementine combines deep domain expertise […] The post Maastricht-based Clementine raises €1.7 million to scale its AI-powered hearing care platform appeared first on EU-Startups .
- GLM-5.3’s Exploits, AI Models and Hardware Speed Up, DeepSeek’s New Agent Harness
The Batch News & Insights: How have software engineering fundamentals changed with agentic coding?
- The New Mid-market Data Center: integrated modernization for AI, Multicloud and Cyber Resilience
The New Mid-market Data Center: integrated modernization for AI, Multicloud and Cyber Resilience IT Pro
- Why a former Nvidia exec is building a Matrix-like model to train AI
Why a former Nvidia exec is building a Matrix-like model to train AI Toronto Star
- Robot from Japan lightens load of restocking New York grocery store
Robot from Japan lightens load of restocking New York grocery store Nikkei Asia
Score: 46🌐 MovesAug 28, 2026https://asia.nikkei.com/business/technology/robot-from-japan-lightens-load-of-restocking-new-york-grocery-store - AI on the edge: 5 reasons to add a deskside AI accelerator like the Dell Pro Max with GB10
AI on the edge: 5 reasons to add a deskside AI accelerator like the Dell Pro Max with GB10 IT Pro
- Imperial and Thomson Reuters pioneer technical approach that could broaden access to sovereign AI
Imperial and Thomson Reuters pioneer technical approach that could broaden access to sovereign AI EurekAlert!
- Meta says its glasses can’t secretly record you anymore, at least not this way
Meta is closing a privacy loophole that allowed smart glasses users to hide the recording indicator while continuing to capture video.
- VURA: A framework for trustworthy AI at scale
“You’re right,” the LLM says. “I was mistaken.” Have you ever read these words during an AI workflow? Nothing kills trust faster than incorrect outputs. It’s no wonder, then, that only a quarter of businesses today fully trust AI to support decision-making and forecasting. And yet, we know AI is business critical. Nine out of 10 businesses are using it; 64% say it’s powering innovation . So, how do you bridge the gap from experimentation to trustworthy deployment? How do you get verifiable, reproducible results from AI at scale? In this article, I’ll show you the framework that’s powering AI success for leading organizations. Why organizations still don’t trust AI We asked 1,400 IT and business leaders what their biggest barriers to success with AI workflows were. One in two (49%) said inaccurate or biased outputs; 38% said it was a reluctance to allow AI to make decisions without human oversight. Then, there was the data issue. Data readiness is an integral part of successful AI workflows. However, half of all organizations said they still faced poor quality or fragmented data. While you don’t need perfect data to start using LLMs, you absolutely need trustworthy data. VURA: The framework for trustworthy AI Closing this trust gap requires two things. First, organizations need a logic layer that connects AI systems to the people who understand the data and business best. Line-of-business teams and analysts cannot sit on the sidelines. They need to help build and validate AI workflows so the logic behind AI’s outputs reflects how the business actually operates. Second, AI workflows and processes should be visible, understandable, repeatable, and auditable. Together, these principles form VURA, a framework we developed to help organizations build and scale trustworthy AI systems. These guidelines will help build trust in your data and your AI’s outputs. You’ll need both if you want your business to build enterprise intelligence. What follows are the four pillars of VURA. Visible Visibility is transparency. Your AI workflows shouldn’t be a black box regarding the data used and the logic applied. Every employee using AI tools should be able to answer two questions: “Where did this answer come from?” and “How did we draw that conclusion?” Otherwise, employees may be working from incorrect information. They could give your customers faulty intel or make important decisions with serious downstream effects. If those answers are still unclear, you may need to tighten your governance or reconsider whether your current AI and data solutions are working. Visibility becomes especially important when AI is used across teams. Understandable It can almost feel like science fiction when tools like ChatGPT or Gemini take the most complicated or vague of prompts, parse through them, and give you an intelligent, thoughtful answer. However, this low threshold for asking and answering virtually any question in natural language isn’t an excuse for glossing over business fundamentals. Your AI systems must be able to explain the logic behind their outputs to even non-technical business users, and your business experts must be able to validate those outputs. Repeatable Repeatable means that with the same AI tools, data, prompts, and business logic, AI will give you the same answer every time. Two people should be able to go to AI with the same question and arrive at the same answer. If an AI system or workflow gives you an excellent answer followed by one that’s clearly wrong, it’s not ready for operationalization. You can’t trust it. Repeatability also requires documentation. When teams identify prompts or processes that help produce reliable outcomes, those should be recorded and shared. Auditable An auditable AI process means you can see what happened. There’s a trail. If there’s an answer or report that seems off, you should be able to identify who owns the workflow, what data and prompts were used, what logic the system followed, and where human judgment and oversight were involved. Auditability is a check and balance for both your AI systems and the human engineers working behind the scenes. Start building trustworthy AI systems today AI can only deliver scalable business value when it’s grounded in trustworthy data and business logic. To operationalize these systems, you’ll have to ensure your AI workflows are visible, understandable, repeatable, and auditable. Alteryx is the transformation and business logic layer that helps you move AI from experimental pilots to trustworthy production. It connects to data wherever it lives, helps business users apply their expertise to AI-powered workflows, and instills the guardrails needed for both your data and your AI systems. With Alteryx, the people closest to the business can shape how data is prepared and applied, while IT gains the governance and auditability required for enterprise use. That’s how AI outcomes become trustworthy. That’s how enterprise intelligence is built. To learn more, visit us here .
Score: 46🌐 MovesAug 28, 2026https://www.cio.com/article/4215176/vura-a-framework-for-trustworthy-ai-at-scale.html - Meta executive leaves for OpenAI as the social media giant faces growing scrutiny in India
Sandhya Devanathan will oversee some OpenAI operations across Southeast Asia and Australia in her new role.
- TASTE: Can AI Models Judge AI Safety Research Proposals?
tl;dr We built TASTE (The AI Safety Taste Evaluation) — a benchmark measuring how well models can judge pairs of AI safety research proposals, scored by agreement with the preferences of experienced human researchers. Two design choices were important for building a high-agreement benchmark (92 pairs, 77% estimated human agreement): a discussion stage in which researchers talk through disagreements before revising their scores, and filtering researchers’ labels for self-reported "strong" confidence. We find models perform worse than human researchers on TASTE (Fable 5, 60%). 📝 Blog , 📄 Paper This work was done as part of the Anthropic Fellows Program . Background While some aspects of AI safety research are relatively straightforward to measure , progress on many questions in AI safety cannot be evaluated with verifiable rewards. For instance, research into mitigating risks from AI misalignment often involves forecasting risks posed by future AI systems. Another example is detecting when models are deceptive , which depends on the difficult task of accurately attributing beliefs and intentions to models. If we want to automate AI safety research — which might become necessary if automated AI research and development outpaces our ability to mitigate the risk of misalignment and misuse — we need reliable measurements of models’ capabilities on the hard-to-verify parts of safety research . One important hard-to-verify aspect of the research process is evaluating research proposals. Judging research proposals well is a high-leverage way to improve research quality, since some proposals are often much more important or tractable than others, and picking a poor initial direction can waste substantial time or resources. In order to evaluate models on the hardest-to-verify tasks, we have to use human judgment as ground truth. However, humans often disagree, making it unclear what the ground truth should be. This is why we focus on finding cases where humans agree and use agreement with humans as our main metric. Building a Research Judgment Benchmark (TASTE) In our paper, we present TASTE (The AI Safety Taste Evaluation), a benchmark of experienced AI safety researchers’ preferences over empirical safety research proposals. Our benchmark contains 92 pairwise comparisons, and we estimate human researcher agreement with our benchmark’s labels at 77%. In general, getting high agreement over fuzzy outputs, such as research proposals, is difficult. A few aspects of our benchmark design are important for improving the quality of human preference labels, measured by agreement with other researchers. We build the benchmark in three stages. First, we use a prompt scaffold with Claude Opus 4.6 to generate AI safety research proposals. Then, we recruit AI safety researchers to rate and comment on the research proposals, reporting their confidence for each set they score. Finally, we filter the dataset of human feedback to retain a high-agreement set of preference pairs. Figure 1: TASTE construction pipeline. Researchers scored each proposal 1–5 on three axes (overall, high-level, approach), ranked the three proposals with ties allowed, and reported a per-prompt confidence. Benchmark preferences are derived from the overall scores. To create the model-written research proposals, we started with a collection of 93 human-written research proposals presented in Anthropic’s Fellows Program. We used Claude Opus 4.6 to reverse-engineer prompts that could motivate each human proposal, and supplemented this set with prompts we wrote ourselves. We generated research proposals using a prompt scaffold that varied paper summaries in context for diversity. See the paper for the full proposal and more details on the scaffold. Figure 2: Example motivating-question prompt and extract from a model-generated research proposal (full proposal in the paper). To construct a human feedback dataset using those research proposals, we asked AI safety researchers to evaluate groups of proposals generated for the same motivating-question prompt. The researchers first gave individual feedback, then discussed disagreements in pairs, and then revised their preferences. Figure 3: Pair-discussion protocol. For each prompt, four researchers gave their feedback individually, discussed disagreements in pairs, then revised their feedback. See the paper for the rubric shown to the researchers. We find that this discussion stage, in combination with filtering for self-reported “strong” confidence, raises estimated human agreement by 15 percentage points, from 53% pre-discussion to 68% for strong-confidence, post-discussion preferences. By looking at discussion transcripts, we find multiple reasons for disagreements among human researchers: substantive disagreements over how well particular techniques would work, and more mundane causes such as one person misreading a proposal text the first time around. We give a fuller breakdown of reasons in the paper. Figure 4: Strong-confidence, post-discussion preferences have higher agreement, and increasing the minimum score gap increases the agreement rate. For each confidence and discussion-stage condition, we compare the preferences from that condition (anchor preferences) with the preferences of the two researchers in the opposing discussion pair (left). Filtering for a gap in “overall score” of at least two points, on a five-point scale, improves estimated agreement for pairs of proposals drawn from the same prompt and pairs drawn from different prompts (right). Error bars show 95% confidence intervals from bootstrap resampling over prompts. To produce TASTE, we take strong-confidence, post-discussion preferences over proposals which may come from the same or different prompts, and filter for pairs where scores differ by at least two points. We further cap the number of times proposals can appear at ten times, for independence between data points, giving 92 pairs with 77% estimated human agreement. To estimate human agreement, we randomly select a researcher’s “overall score” per proposal from the opposing discussion pair, and take the proposal with the higher sampled score as preferred. This allows us to evaluate pairs of proposals where no single researcher rated both of them. This does not perfectly capture human performance, and instead approximates "I assign a different person to score each proposal; the ‘preferred’ proposal is the one that receives the higher score". A more typical measure of inter-rater agreement – comparing only the pairs where another researcher scored both proposals – gives 83% agreement but only validates 50 of the 92 pairs. Evaluating Models’ Research Judgment We measure model performance on TASTE, comparing against the human labels. We use two setups: a standard setup where the model sees two proposals in context and gives a probability to each proposal winning which we binarize to give a preference, and a tougher single-proposal scoring setup where the model sees each proposal individually in context, scores it, and is assessed on the implied preferences from its scores against the preferences in TASTE. In the standard setup, we find that the best model performs worse than our human researchers — Fable 5 achieves 60% whereas we estimate researcher performance at 77%. While there is currently a noticeable gap versus human performance, we think future models could close it: Fable 5 achieves 69% accuracy on 74 pairs of proposals drawn from different prompts, and we find some evidence that models over-focus on how well proposals answer the motivating question when shown on pairs from the same prompt. See the paper for more details. Figure 5: Model performance on TASTE by release date and model provider. Fable 5 achieves 60%, and almost all models perform within 2 standard deviations of chance. Opus 5 and GPT-5.6-Sol perform near chance on TASTE despite being at the frontier on general agentic benchmarks. Per-model confidence intervals span roughly ±10 percentage points, as we have a limited number of preference pairs making it difficult to draw conclusions about relative model performance. Conclusion As AI capabilities improve, it might be necessary to automate AI safety research. We built a benchmark measuring models’ research judgment through their ability to choose between AI safety research proposals. Our results show model performance still trails that of expert human researchers. More diverse and larger-scale evaluations of models’ AI safety research capabilities will be needed going forward. Our findings suggest that future data-collection efforts can improve human label quality by including a pair-discussion stage and by filtering labels for self-reported confidence. To learn more, read our paper . We are sharing TASTE with AI safety researchers. For access, please fill out this form . Discuss
Score: 46🌐 MovesAug 28, 2026https://www.lesswrong.com/posts/iSDbyrG8yfqk3KJbT/taste-can-ai-models-judge-ai-safety-research-proposals - Meta smart glasses ban trend is catching steam in the US, and an NY nightclub just set the ball rolling
A New York City nightclub has banned smart glasses as concerns over discreet recording continue to spread, following similar restrictions across the UK.
- The Sequence Robotics - Issue #922: Learning About LeRobot: The Transformers Moment for Robots
Robotics Gets Its PyTorch Stack.
Score: 46🌐 MovesAug 28, 2026https://thesequence.substack.com/p/the-sequence-robotics-issue-922-learning - How ChatGPT Justifies Its CPMs; Is “Chatbot Ads” Its Own Specialty?
Splashy new programmatic media sellers always seem to want an outlandish $60 CPM. And: ChatGPT doesn’t seem to have nailed down its ideal advertiser just yet. The post How ChatGPT Justifies Its CPMs; Is “Chatbot Ads” Its Own Specialty? appeared first on AdExchanger .
- An AI coding tool whose underlying model advertises a million-token window silently had its context cap cut by 27 percent overnight, and the new ceiling sits exactly where the bill doubles
A coder opens a project and everything looks normal. The files load, the session starts. Then, midway through a multi-file rewrite, the tool hits a wall and the errors start stacking. No warning arrived. No update landed. Yet the tool that handled the whole codebase yesterday cannot finish the job today. The question is not ... Read more
- Chinese spatial-intelligence firm Manycore launches Lux3D and starts world-model testing
Manycore Tech, a Chinese spatial-intelligence company, launched Lux3D on Aug. 27 and began testing a next-generation world-model product. Lux3D generates 3D assets from text or images, with its Turbo Mode producing models in as little as 20 seconds. The company has also opened API access, allowing batch generation through Harness Mode for workflows including e-commerce, […]