AI News Archive: May 29, 2026 — Part 4
Sourced from 500+ daily AI sources, scored by relevance.
- Qilimanjaro Pushes Analog Quantum as AI Compute Demands Surge
Qilimanjaro says analog quantum systems could reduce error correction and accelerate AI, optimization, and simulation. On May 28, its analog system joined the digital quantum computer at the Barcelona Supercomputing Center. The post Qilimanjaro Pushes Analog Quantum as AI Compute Demands Surge appeared first on EE Times .
Score: 47🌐 MovesMay 29, 2026https://www.eetimes.com/qilimanjaro-pushes-analog-quantum-as-ai-compute-demands-surge/ - Workers need greater say over AI rollout, says TUC-backed report
Exclusive: IPPR thinktank calls for new measures to boost employees’ influence at ‘pivotal moment’ in history Workers urgently need more bargaining power over the way AI is adopted in the workplace to ensure the benefits are fairly shared, according to a TUC-backed report from a leading thinktank. The Institute for Public Policy Research (IPPR) is calling for a package of measures to boost employees’ influence at what it calls a “pivotal moment in the history of work”. Continue reading...
- Who will win in an autonomous future?
Who will win in an autonomous future? Automotive News
Score: 47🌐 MovesMay 29, 2026https://www.autonews.com/resource_center/roland-berger/an-who-will-win-in-an-autonomous-future/ - Verifying Agentic Development at Scale
What we've learned building end-to-end testing capabilities in Devin's virtual machine
- OpenAI Flexes Enterprise Ambitions With Colin Fleming As Business CMO
The senior marketing hire reflects OpenAI’s plans to compete as a leading enterprise B2B platform and brand on par with the top global technology and services firms.
- India Leads Global AI Adoption—But Faces the Highest ‘Complexity Cost’
Freshworks Inc. today released The Global Cost of Complexity Report: The Mid-Market AI Complexity Trap, featuring a detailed examination of how AI complexity is affecting mid-market companies in India. The research, based on responses from over 9,000 mid-market IT decision-makers across six countries, finds that in India, 27% of the average mid-market AI budget is lost to […] The post India Leads Global AI Adoption—But Faces the Highest ‘Complexity Cost’ appeared first on CXOToday.com .
- Weekly Rundown—Moffitt Cancer Center expands Reimagine Care's virtual oncology model; Tanner Health deploys AI workforce solution
Stay up to date on the latest in health tech, digital health and health AI news with this weekly brief.
- Microsoft warns GPU mining malware is being spread to users through SEO poisoning and AI chatbots — cryptojacking campaign targets gamers and high-end PC users with downloads disguised as popular PC utilities
Microsoft has uncovered a GPU-focused cryptojacking campaign that used SEO poisoning and, in some cases, AI chatbot software recommendations to spread malware disguised as popular PC utilities like HWMonitor and CrystalDiskInfo
- Musk defends AI ambitions as IPO reveals trouble
Elon Musk insists that his artificial intelligence venture xAI remains a serious competitor, pushing back against mounting doubts after revelations that the supercomputing facilities built to power his own AI models are being rented out to a rival.
- Picogrid CEO Zane Mountcastle on new opportunities after $45 million funding
Picogrid CEO Zane Mountcastle on new opportunities after $45 million funding
- Beyond Serendipity: Redefining Rational Molecular Glue Discovery via AI-driven Methodologies
Beyond Serendipity: Redefining Rational Molecular Glue Discovery via AI-driven Methodologies azcentral.com and The Arizona Republic
- Maharashtra will provide innovators access to 2,000 GPUs: CM Devendra Fadnavis
Fadnavis also talked about plans to attract ₹10,000 crore in investments, create 125 lakh jobs and build dedicated infrastructure to support start-ups and innovation in the sector, under its recent AI policy.
- AI Used to Be Generative. Now It’s All About Agents
Corporate speak has changed when it comes to describing the evolving technology
- Peak XV-backed C2i Semiconductors secures funding in extended Series A round
Peak XV-backed C2i Semiconductors secures funding in extended Series A round DealStreetAsia
Score: 45💰 MoneyMay 29, 2026https://www.dealstreetasia.com/stories/c2i-semiconductors-funding-483951 - Labcorp, Amazon Web Services discuss AI benefits, even as guardrails crucial
Executives from Labcorp and Amazon Web Services outlined artificial intelligence uses at Triad BioNight, from speeding diagnoses to improving efficiency while maintaining strict data safeguards.
- PwC India, Leah Expand Alliance to Drive Enterprise AI Adoption
Partnership focuses on agentic AI deployments across finance, HR, procurement and enterprise ops.
Score: 45🌐 MovesMay 29, 2026https://analyticsindiamag.com/ai-news/pwc-india-leah-expand-alliance-to-drive-enterprise-ai-adoption - AWS reportedly to tuck Elon Musk's Grok into Bedrock, despite zero enterprise demand
The energy drink of frontier models
- Europe Cannot Afford to Miss the Intelligence Era
Europe Cannot Afford to Miss the Intelligence Era uk.entrepreneur.com
Score: 45🌐 MovesMay 29, 2026https://uk.entrepreneur.com/technology/palantir-london-ai-sovereignty-europe - India emerges as a key Codex market with 27x user growth: OpenAI
India emerges as a key Codex market with 27x user growth: OpenAI
- Micron Faces New Threat From Samsung’s Memory Chip for AI
Micron Faces New Threat From Samsung’s Memory Chip for AI Barron's
Score: 45🌐 MovesMay 29, 2026https://www.barrons.com/articles/micron-stock-price-samsung-memorychip-ai-ac9a8e59 - Shark Tank Star Is Fighting Phantom Bots While Utah Locals Fight His Data Center
Seems like people just don't like your project, dude.
Score: 45🌐 MovesMay 29, 2026https://gizmodo.com/shark-tank-star-is-fighting-phantom-bots-while-utah-locals-fight-his-data-center-2000765137 - B.C. Greens calling on province to stop AI data centres, contain water usage
B.C. Greens calling on province to stop AI data centres, contain water usage CBC
- BYD's new 4-nm self-driving chip fails to dispel investors' growth concerns
BYD's new 4-nm self-driving chip fails to dispel investors' growth concerns Nikkei Asia
- AI Dark Output: The Visible Cost of Invisible Output
Why AI's increasing output is going to be one of the hardest economic measurement problems in history. AI "Dark Output" could end up being the majority of economic activity, but a challenge to measure
Score: 45🌐 MovesMay 29, 2026https://newsletter.semianalysis.com/p/ai-dark-output-the-visible-cost-of - CrowdStrike Expands Project QuiltWorks with Cyber Insurance Industry Leaders to Combat Financial Exposure to Frontier AI Risk
CrowdStrike announced the next evolution of Project QuiltWorks, extending the industry framework from securing frontier AI risk to mitigating financial exposure. As frontier AI accelerates vulnerability discovery and compresses exploitation timelines, leaders from the cyber insurance industry – Coalition, Liberty Mutual Insurance, Lockton, Resilience, and Marsh – bring actuarial intelligence, underwriting expertise, and financial protection to the […] The post CrowdStrike Expands Project QuiltWorks with Cyber Insurance Industry Leaders to Combat Financial Exposure to Frontier AI Risk appeared first on CXOToday.com .
- Shadow AI: The hidden risk expanding across the enterprise
Companies and employees are racing to capture the value and efficiencies offered by AI, but security is often an afterthought. Employees are using unauthorized GenAI tools to summarize documents, draft emails, and analyze potentially sensitive or proprietary data. Developers are adding AI capabilities before security teams can review them. SaaS platforms are adding AI features that may process sensitive business data by default. The result is a new attack surface expanding faster than most organizations can govern. For CISOs and CIOs, the challenge is twofold. You must secure how employees use AI in daily work, and you must protect the AI-enabled applications your organization is building and consuming. Without visibility across both, shadow AI becomes a blind spot where data can move, policies can fail, and adversaries can operate with less resistance. Shadow AI is bigger than unauthorized chatbots Shadow AI goes beyond employees pasting content into public chatbots. It includes unapproved AI assistants, embedded copilots inside SaaS applications, unapproved AI features, and internally developed AI workflows that bypass governance. Many organizations lack a unified view of where AI is being used, the data being exposed, or where or how to apply controls. Security teams are left unable to answer basic, yet critical, questions: Which AI services are employees accessing? What sensitive data is being shared? Are developers connecting proprietary code or customer data to external models? As the uncertainty increases, so do the risks of data leakage, compliance failures, inconsistent policy enforcement, and reputational damage. AI-native threats are already here Enterprises face new AI-specific attacks. For example, prompt injection techniques can manipulate models into exposing information, ignoring safeguards, or taking unintended actions. Indirect prompt injection is especially dangerous because malicious instructions may be hidden in trusted sources such as documents, websites, or knowledge bases. Prompt injection is a broad and rapidly evolving threat landscape that warrants dedicated attention. For a deeper exploration of how these attacks are defined and categorized, we recommend reviewing our comprehensive overview: Prompt Injection: Definition and Attack Taxonomy. Why traditional security falls short Traditional security tools were built for a different era defined by network perimeters, known attack signatures, and human-driven interactions. They were never designed to interpret the intent or content of AI interactions. Web proxies and firewalls cannot inspect encrypted traffic. Locally running AI applications may operate entirely on the endpoint and generate no network telemetry. Zero Trust and network segmentation, while foundational to modern security strategies, were built around human-to-system interactions — not the emerging reality of agent-to-agent and agent-to-tool communications, where autonomous AI systems make access decisions at machine speed, outside the reach of traditional policy enforcement. Perhaps most importantly, while Zero Trust can govern which data a user is permitted to access directly, it cannot control which data becomes accessible through an LLM via retrieval, tool calls, or agentic workflows acting on the user’s behalf. That is a fundamentally different problem, and one that conventional architectures were never designed to solve. The result is a dangerous gap between existing security coverage and emerging AI risk. Organizations may have strong controls across endpoint, identity, and cloud, and still miss the moment sensitive data is exposed through a GenAI tool, or when an AI workflow is manipulated through malicious input. Closing that gap requires a purpose-built approach. CrowdStrike Falcon® AI Detection and Response (AIDR) is designed to provide the visibility, control, and protection that AI-driven environments demand. It can identify and stop AI-specific threats such as prompt injection, data leakage, and credential abuse targeting AI services, before they become breaches. Where traditional tools see infrastructure, CrowdStrike sees the full picture: which AI is being used, which data and prompts are reaching those systems, and whether the interactions represent risk. By unifying protection across endpoint, identity, cloud, and AI on a single platform, CrowdStrike enables security teams to defend AI-powered applications with confidence and reduce risk without slowing the business. 3 actions to take now First, assess shadow AI exposure by identifying which AI tools are in use, where AI features are enabled in SaaS applications, and which sensitive data is already flowing to those services. Second, define governance that matches real usage. Establish approved tools, acceptable use policies, and review processes for AI applications and integrations before they reach production. Third, deploy integrated controls to prevent access or data egress to unauthorized AI services, detect prompt injection and AI-related abuse, and monitor for adversary activity across identity, cloud, and endpoint. Turn AI into an advantage AI creates real business value, but without visibility and control, it expands the attack surface in ways traditional security wasn’t built to handle. Shadow AI cannot be left unmanaged, and fragmented tools cannot keep pace with how quickly AI is being adopted across the enterprise. CrowdStrike unifies AI visibility, control, and protection on a single platform built for how AI is used in the modern enterprise. Security teams gain the insight they need, and the business keeps moving. To learn more about CrowdStrike, visit here .
Score: 45🌐 MovesMay 29, 2026https://www.cio.com/article/4178708/shadow-ai-the-hidden-risk-expanding-across-the-enterprise.html - Gartner Says CFOs Gain a Competitive Advantages from Strategic AI Deployment, Not AI Spending Levels
Gartner Says CFOs Gain a Competitive Advantages from Strategic AI Deployment, Not AI Spending Levels Gartner
- NVIDIA Introduces X-Token: Projection-Guided Cross-Tokenizer KD That Outperforms GOLD by +3.82 Average Points on Llama-3.2-1B
NVIDIA Introduces X-Token: Projection-Guided Cross-Tokenizer KD That Outperforms GOLD by +3.82 Average Points on Llama-3.2-1B MarkTechPost
- MeMo's memory model lets teams upgrade their LLM without retraining it — and performance jumps 26%
Enabling LLMs to acquire new knowledge after training remains a major hurdle for enterprise AI — current solutions are either too expensive, too slow, or constrained by context window limits. MeMo , a framework from researchers at multiple universities, encodes new knowledge into a dedicated smaller memory model that operates separately from the main LLM. The modular architecture works with both open- and closed-source models and sidesteps the complexity of RAG pipelines and full model retraining. Experiments show that MeMo handles complex queries reliably even when retrieval pipelines are noisy. It avoids the catastrophic forgetting associated with direct fine-tuning and provides a cost-effective pathway for continuous knowledge updates. The challenge of updating LLM memory Large language models are frozen after training and their internal knowledge remains static until they undergo subsequent, computationally massive updates. Currently, developers rely on three main approaches to integrate external knowledge into an LLM, each with distinct drawbacks: Non-parametric methods , such as retrieval-augmented generation (RAG) and in-context learning , retrieve relevant documents from an external database and insert them directly into the model's prompt. While popular, these methods are limited by context window sizes. As Armando Solar-Lezama, a co-author of the paper, told VentureBeat, “Vector databases have a fundamentally difficult job of encoding the full semantics of a chunk of text in a single vector, and then match that vector to a query, even when the relevance of the chunk... may only be apparent in the context of other chunks.” The researchers note that the semantic similarity of embeddings often does not correspond to what a user's query actually requires. Processing thousands of retrieved tokens also creates substantial computational overhead and inference latency. Most problematically, RAG systems are highly sensitive to noise. Irrelevant or poorly retrieved passages often degrade the model's final response. Parametric methods , like continual pretraining or supervised fine-tuning, attempt to internalize new knowledge directly into the LLM's weights. Updating modern, massive LLMs is prohibitively expensive and typically impossible for proprietary, closed-source models hidden behind APIs. Fine-tuning is also prone to causing catastrophic forgetting . Forcing the model to adapt to new corporate data often erodes its previously acquired reasoning capabilities and safety guardrails. Latent memory methods , such as context compression, offer a middle ground. They compress knowledge into compact "soft tokens" or representations that are added to the model’s context during inference. The fatal flaw here is "representation coupling." The compressed memory is strictly bound to the model architecture that produced it; you can't transfer a latent memory trained on an open-source model to a closed-source one. How MeMo works The MeMo (Memory as a Model) framework introduces a modular architecture featuring two separate components. The MEMORY model is a small language model trained specifically to encode new knowledge into its parameters. The EXECUTIVE model is a frozen, off-the-shelf LLM that functions as the reasoning engine. When a user asks a question, the EXECUTIVE model treats the MEMORY model as an external oracle, issuing targeted sub-queries to gather facts and synthesizing those facts into a final answer. The core design principle driving MeMo is the concept of "reflections." Reflections are targeted question-answer (QA) pairs designed to capture every possible angle of a knowledge corpus. Rather than forcing the AI to process a massive, unstructured document corpus during training, MeMo uses a GENERATOR model to distill the raw text into thousands of targeted QA pairs. The MEMORY model is then fine-tuned on this dataset to answer questions using only its parametric knowledge without the need to read retrieved context. At inference time, the interaction between the two models follows a structured, three-stage protocol: 1. The EXECUTIVE model decomposes a user's complex query into a set of atomic sub-questions. The MEMORY model answers each independently to establish the basic facts. 2. Using those initial clues, the EXECUTIVE model issues follow-up queries to narrow down candidate entities until it confidently converges on a specific target. 3. Finally, the EXECUTIVE model queries the MEMORY model for supporting facts about that target entity and synthesizes the retrieved snippets into a cohesive answer. This architecture merges the strengths of the three existing AI memory paradigms while bypassing their pitfalls. It leverages off-the-shelf frontier models by keeping memory storage separate from reasoning, guaranteeing compatibility with both open-weight and closed API models. It internalizes knowledge directly into parameters, but isolates the updates to a smaller, dedicated MEMORY model to protect the reasoning engine. Finally, it creates a queryable memory artifact that is not tied to any specific model and can be used with different LLM families. Handling continual knowledge updates Managing an AI's memory requires continuous updates as company policies change and new reports are published. Normally, updating a model's parameters requires retraining it from scratch on both the old and the new data combined. As the knowledge base grows, this cumulative retraining cost becomes unmanageable. To handle continual updates efficiently, MeMo relies on a technique called "model merging." Instead of a massive joint retraining phase, MeMo trains a new, independent MEMORY model exclusively on the newly added documents. The system derives a "task vector" representing the parameter changes learned from the fresh data. These updates are then mathematically merged into the weights of the original MEMORY model. This approach reduces the computing hours required to keep the system current while avoiding the interference that causes catastrophic forgetting. This efficiency comes with a trade-off: model merging incurs an 11% to 19% accuracy drop compared to a full retrain, depending on the reasoning model used. MeMo in action To measure real-world effectiveness, the research team evaluated MeMo against several industry benchmarks that require complex, multi-hop reasoning across multiple documents. The researchers used Qwen2.5-32B-Instruct as the GENERATOR model to distill raw text into reflections. For the primary MEMORY model, they deployed Qwen2.5-14B-Instruct. They also validated the approach on smaller 1-2B parameter models across different architectures, including Gemma3-1B. For the EXECUTIVE reasoning model, they tested both the open-weight Qwen2.5-32B and Google's proprietary Gemini 3 Flash. They benchmarked MeMo against a "Perfect Retrieval" upper bound (where the exact correct documents are manually provided) and several advanced retrieval systems, including traditional BM25 search, dense vector retrieval, and state-of-the-art graph-based RAG (HippoRAG2). They also tested "Cartridges," a recent method that loads a trained KV-cache onto the model during inference. MeMo dominated in long-document reasoning. On the NarrativeQA benchmark, MeMo achieved 53.58% accuracy paired with Gemini 3 Flash, according to the researchers. HippoRAG2 maxed out at 23.21%. Enterprise systems frequently need to synthesize complex answers, such as traversing overlapping regulatory frameworks written independently by different bodies, or consolidating insights across a massive codebase and external documentation. Traditional RAG systems falter here because they hit context window limits and fail to connect concepts spanning hundreds of pages. MeMo succeeds because those connections are mapped and internalized inside the MEMORY model during training. It is "like having your very own Malcolm Gladwell that can connect the story of the Beatles with the story of Bill Gates to make an argument about the nature of expertise," Solar-Lezama said. The experiments revealed another major advantage: upgrading the reasoning engine requires zero retraining. Simply switching the EXECUTIVE model from the open-source Qwen to the proprietary Gemini 3 Flash boosted MeMo's performance by 26.73% on NarrativeQA and 11.90% on the MuSiQue benchmark. For practitioners, this means you can train a MEMORY model securely on your private data and instantly plug it into the latest commercial APIs, continuously upgrading system intelligence without incurring new training costs. The research team described the integration as requiring no additional setup: "The base (or Executive) LLM that teams are already using in RAG can be configured to query the Memory model directly. These queries are done in natural language, similar to sending a message request to an API, with no additional setup required." MeMo also handles noisy data exceptionally well. When researchers deliberately flooded the dataset with irrelevant documents (up to twice the amount of the useful information), HippoRAG2’s performance dropped by 11.55%. MeMo's performance remained relatively stable, dropping less than 2%. Enterprise knowledge bases are typically messy, filled with duplicate documents and outdated policies. Standard RAG systems struggle with this noise, pulling incorrect paragraphs into the prompt and causing hallucinations. Because MeMo's EXECUTIVE model interacts with a synthesized oracle rather than raw document chunks, it remains highly robust against disorganized corporate data. Limitations and trade-offs For engineering teams looking to deploy MeMo, there are several key limitations to consider. Unlike traditional RAG systems that quickly index raw documents into a vector database, MeMo requires an upfront training cost for each new corpus. The data generation pipeline used to synthesize the training reflections is computationally expensive. For example, the team noted that "generating the full reflection QA dataset took approximately 240 GPU-hours on NVIDIA H200s," while training a 14B parameter MEMORY model "took approximately 180 H200 GPU-hours." As Solar-Lezama said, "Reducing the training cost is one of the most significant open research problems in order to make this a workhorse technique." Because the MEMORY model is a fixed-size neural network, its ability to internalize knowledge is bounded by its representational capacity. While the researchers did not hit a hard limit during their benchmarking, they hypothesize that “sufficiently large or information-dense corpora will exceed what a fixed-size MEMORY model can correctly compress and represent.” Finally, because MeMo synthesizes answers from parametric memory rather than retrieving exact text snippets, it obscures the provenance of the information. This makes it difficult to attribute specific claims to original source documents, which poses a critical compliance issue for enterprise applications requiring strict audit trails. Deciding between MeMo and traditional RAG comes down to a heuristic of "lookup vs. synthesis," alongside data volatility. The researchers advise that "traditional RAG would be preferred when answers live in a single document or when there is a well-defined source... MeMo would be preferred when the task shifts from lookup to synthesizing an answer from information scattered across multiple chunks." If your knowledge corpus changes rapidly (e.g., daily feeds) and you require exact source citations, RAG remains the better option due to the upfront training cost of MeMo. If your corpus consists of generalized domain knowledge that evolves slowly relative to its volume, MeMo offers vastly superior reasoning. Teams can also adopt a hybrid routing architecture in production: sending "lookup" queries to a standard vector database and "synthesis" queries to the MEMORY model. "Looking further out, I would expect memory models to become a standard architectural component alongside retrieval," Daniela Rus, co-author of the paper and director of the MIT Computer Science and Artificial Intelligence Lab (CSAIL), told VentureBeat, "in the same way that caching and indexing are standard components of any serious data system today."
Score: 45🤖 ModelsMay 29, 2026https://venturebeat.com/orchestration/memo-memory-model-teams-upgrade-llm-without-retraining - AI search may kill the click. But users still need to trust the answers
It’s hard to think of Google as a comeback story, because it never went away, but in the world of AI , it’s a fitting narrative. The company was broadly mocked for its early moves in the generative era, which were mostly stumbles . Upstarts like Perplexity and ChatGPT were nipping at its market share with arguably more innovative experiences. It had to navigate choppy antitrust waters. Today, Google is in a much stronger position. Not because it’s just coming off its I/O developer conference , or because it has the best model or because it commands a major AI ecosystem. Those titles get passed around the big labs every few months. The reason is much simpler: the business is holding up. AI hasn’t broken Google Alphabet’s first-quarter earnings showed Google Services revenue up 16% to $89.6 billion and Google Search and “Other” revenue is up 19%. Clearly, the rising presence of AI in the information ecosystem hasn’t hurt Google’s business; if anything, it’s the opposite. That success appears to have led to more confidence. Google announced many new AI products at I/O, but one of the most notable ones to the media industry was a set of new ad formats . Conversational Discovery ads are built on the fly to fit naturally into the answer to the person’s query, appearing as a “sponsored” section. Highlighted Ads and AI-powered Shopping Ads are similar, inserting ads into more general product category-specific queries. And then there are Business Agents for Leads—tailored versions of Gemini that appear within the ad. These formats are still being tested, but the direction is clear: Google is getting more sophisticated about how it monetizes AI experiences. The company stated a few months back that it had no plans to sell ads in Gemini, which executives mentioned in response to ChatGPT ads. And, sure, Google can still say that the Gemini chatbot is not becoming an ad product. But that distinction feels less meaningful now that so many Gemini-powered AI experiences across Search are being commercialized. Of course, all those AI-powered ads appear within or next to an answer. And that answer is supposed to be made of the best information Google can find, which is often from media publishers. In the old system, Google sold ads that were prominent in results, and those ads benefitted from the close proximity to links from trusted media sources. Google the best SUVs, and you may see ads for Toyota or Hyundai before you see a link to Car and Driver. Now the information, built in part from the publisher’s content, is right there on the result. The user gets the info, the AI-powered ad provides a path to transact, and everything is handled without any need for them to ever leave Google. Instead of monetizing the path to information, Google is now monetizing the information experience itself. The party left out of that bargain is of course the publisher. In many cases, their content was the raw material that informed the answer. When AI search was relatively new, Google would claim—truthfully—that the audience that visits a publisher site from AI answers is more likely to engage and transact. But why would they when Google is providing the means to do that before they ever arrive? These latest ad experiences seemingly point to a bad situation getting worse. Why sources still matter However, there are layers to this. Users don’t care about business models; whether or not they have an inclination to buy something or engage depends on not just the content of the answer but how much they trust it. A study published in Nature described trust in AI as dynamic and context-dependent. In other words, it changes depending on the nature of the AI experience and over time. And another study by the Reuters Institute found users had moderate trust in AI answers, but they also value their speed and aggregation. So yes, people like the utility of AI, but trust is conditional. And one of the most important assets any media brand has is the trust it cultivates over time. Imagine two AI answers about the same product: one built from social posts, blogs, Reddit threads, and online forums, and the other built from articles on Consumer Reports, the Wirecutter, Time, and CNET. Which one sounds more trustworthy? In short, citations matter. People will be more inclined to trust answers created from brands that they’re familiar with. While there is little data about AI ad experiences directly, the entire media ad model is founded on this idea—that an ad doesn’t just benefit from being present on a platform but also by being associated with that platform’s brand. Google has so far not been that concerned with what publishers want. But Google does need advertisers to believe AI search ads work. If advertisers see better performance when ads appear beside credible, well-sourced answers, they will care about the quality of those answers. That could create pressure on Google to maintain a healthier source ecosystem. That pressure may not look like simple licensing deals. It could involve clearer traffic paths, richer citation treatment, new publisher products, commercial partnerships, or advertiser demand for premium source environments inside AI search results. Moving on from the click Review sites are the clearest example because the transaction is obvious. If someone asks for the best dishwasher, the AI answer can cite reviews and then push the user toward purchase. But the same logic applies beyond commerce: A health answer, a travel plan, or even a summary of a political issue all depend on source trust. Even when there’s no immediate checkout, the user’s confidence in the answer shapes what they believe and what they do next. The warning is clear: Google’s new push into AI ad experiences could further weaken traditional publisher revenue streams, especially traffic-based display, affiliate, and search-driven monetization. But there’s another side to the equation: If AI answers need credibility to be useful, then credible media still has value. That value may no longer show up as a click. But it will still shape whether users trust the answer enough to act on it.
- How AI can be employed in emergency response
How AI can be employed in emergency response Healthcare IT News
Score: 44🌐 MovesMay 29, 2026https://www.healthcareitnews.com/video/how-ai-can-be-employed-emergency-response - Gartner Says Supply Chain Confront Geopolitical and AI Challenges
Gartner Supply Chain Symposium highlights strategies to navigate chaos, orchestrate agility, and accelerate Innovation. The post Gartner Says Supply Chain Confront Geopolitical and AI Challenges appeared first on EE Times .
Score: 44🌐 MovesMay 29, 2026https://www.eetimes.com/gartner-says-supply-chain-confront-geopolitical-and-ai-challenges/ - BYD Offers Crash Cost Coverage for God’s Eye Assisted Driving
BYD says it will cover some God’s Eye crash costs, but drivers still need to know when the guarantee applies and where the limits are. The post BYD Offers Crash Cost Coverage for God’s Eye Assisted Driving appeared first on TechRepublic .
Score: 43🌐 MovesMay 29, 2026https://www.techrepublic.com/article/news-byd-gods-eye-crash-cost-coverage-apac/ - AI Computer Design Diversity A Boon For Synopsys, Says CFO
AI companies are increasingly turning to custom computer designs to optimize their systems, which should help Synopsys stock. The post AI Computer Design Diversity A Boon For Synopsys, Says CFO appeared first on Investor's Business Daily .
Score: 43🌐 MovesMay 29, 2026https://www.investors.com/news/technology/synopsys-stock-tailwind-heterogeneous-ai-systems/ - From Billions Of Violations To Actionable Insights: Calibre Vision AI
Deliver faster insight, eliminate wasted debug cycles, and significantly compress overall DRC iteration time at advanced nodes. The post From Billions Of Violations To Actionable Insights: Calibre Vision AI appeared first on Semiconductor Engineering .
Score: 43🌐 MovesMay 29, 2026https://semiengineering.com/from-billions-of-violations-to-actionable-insights-calibre-vision-ai/ - New Silicon Motion SM2524XT chip brings 14 GB/s to mainstream SSDs — 6nm DRAMless controller boasts heavy AI PC optimization and slashes KV cache latency
Silicon Motion announces its SM2524XT mainstream SSD controller that promises 14 GB/s read speed, up to 2.5 million random IOPS, and sustained random performance.
- Protecting against inference theft
HTTP requests are inexpensive. Vercel charges ~$2/million, a fraction of a cent per call. But a single prompt to an agent on a frontier model can cost $2, making AI a million times more expensive, and inference theft one of the highest-margin businesses an attacker can run. We have seen this type of attack on our own APIs. If you have AI endpoints exposed to the internet, the risk of abuse is high and can easily run up bills in the tens of thousands of dollars or more. Protecting those endpoints requires verification to run on every AI request, not on the session or signup. Rate limits and auth walls aren't sufficient on their own because checks that run once per session get amortized away across thousands of stolen calls. At Vercel, we gate every AI request through BotID deep analysis, and you can do the same on your own endpoints with a few lines of code. What inference theft is Inference theft is the unauthorized use of someone else's paid AI inference, either for free consumption or downstream resale. The operator pays per AI call; the attacker pays nothing for the inference, then resells the tokens at a discount. This goes beyond rate-limit abuse to actual resale of a stolen resource in a market. Which AI endpoints are at risk? Any internet-facing endpoint that gives a caller meaningful control over an LLM prompt is a target. The more general the endpoint, the higher the payout per stolen call. AI playgrounds, like the AI SDK Playground , are the most dangerous shape because the caller has maximum control over the prompt, the model, and often the parameters. Stolen calls land cleanly into any standard client. Support bots and documentation assistants are less exposed when system prompts are fixed server-side, but attackers have learned how to talk the models around system prompts cheaply enough to make resale viable. Resale value tracks how easily the stolen calls can be dropped into a provider-compatible client. Why web defenses don't mitigate inference theft IP rate limits and auth walls were built for attacks with dramatically lower per-call economics, where gaming IPs and accounts weren't worth the cost. The payoff from stolen inference is high enough that attackers will procure residential proxy IPs by the thousands and register throwaway accounts at whatever scale defeats your gate. Rate limits get diluted across the fleet of IP addresses, and real accounts pass authentication. The architecture of abuse Sophisticated attackers wrap your custom AI endpoint in an OpenAI- or Anthropic-compatible adapter and fan calls out through residential proxies. The adapter is the key component. It is a one-time engineering cost that presents the victim's idiosyncratic API as OpenAI- or Anthropic-compatible, so the stolen inference drops into any standard coding agent or SDK. Resale at even five to ten percent of list price against zero marginal inference cost can make for a generous-margin business. A recent example is Chipotlai Max , a forked coding agent that ships with a proxy turning Chipotle's customer-support chatbot into an OpenAI-compatible endpoint. The project openly solicits help porting the same inference theft approach to Home Depot, Lowe's, Target, and Starbucks. The adapter is also the session boundary for the attacker's downstream users. They authenticate to the adapter, not to your endpoint. By the time a call hits your API, it has already crossed the boundary you were planning to defend. The check has to run on the call the adapter is proxying, not the session it sits behind. The shape of a real attack on our own endpoint On April 12, 2026, traffic to the Vercel docs AI chat endpoint spiked to roughly ten times normal volume on Anthropic's Claude Haiku 4.5 model. Traffic rose to 1,300 requests per minute at peak, which would have translated to an inference cost run rate of over ten thousand dollars per day. The attack came in through residential proxies that obscured the real client IPs. Across hundreds of thousands of bot requests over two days, standard per-IP rate limits had nothing useful to act on. How to defend against inference theft Protecting AI endpoints against inference theft requires verification of every request. We use Vercel's BotID with deep analysis, called inside the route handler before the AI request lands. Verification has to run on every AI request If our gate had run at session start instead of per request, the attacker would have paid the bypass cost once and walked away with hundreds of thousands of stolen calls. Any check that runs per session amortizes the attacker's bypass cost across every subsequent inference call. Per-request gates force that ratio down to one, and even at high inference prices, defeating a check on every call isn't worth the cost. This is where the cost asymmetry works in the defender's favor. Inference is the most expensive resource per call the attacker is stealing, but verification is one of the cheapest costs per call for protection. Implementing request verification with BotID deep analysis Traditional image CAPTCHAs no longer hold up against modern attackers because the same AI models that make inference worth stealing can easily bypass them. We deploy Vercel BotID on our AI endpoints, gating every request. BotID is an invisible CAPTCHA with deep analysis powered by Kasada that uses client-side machine learning to distinguish humans from bots without showing a visible challenge, which means it can run on every request rather than only at session start. BotID deep analysis detected and blocked more than ten thousand bot requests in the first minutes of the spike. Within twenty-four hours, request volume on the endpoint was flat at normal levels. Server-side, checkBotId() runs inside the route handler and returns a classification for the request currently being served. The route also has to be declared on the client. Without this, checkBotId() fails because BotID doesn't attach the challenge headers to the request: See the BotID docs for the next.config.ts wrapper and the full setup. Protect inference, not just access Inference will stay orders of magnitude more expensive than the requests carrying it, so resale stays profitable and attackers will keep iterating. To protect your AI endpoints: Audit which of your AI endpoints are exposed Prioritize by attack likelihood: more caller prompt control means an easier target Gate every endpoint on every request Get started in our AI endpoint protection Knowledge Base Guide . Read more
- DeepSeek makes price cuts permanent - Asian Tech Roundup
DeepSeek makes price cuts permanent - Asian Tech Roundup Computing UK
Score: 43🌐 MovesMay 29, 2026https://www.computing.co.uk/news/2026/ai/deepseek-makes-price-cuts-permanent-asian-tech-roundup - The Chatbots and Agents Are Going To Merge
Could the AI labs intuit your next move and accomplish it for you? That seems to be where they're heading.
- SentinelOne Stock Craters After Earnings. It’s Cutting Jobs for AI.
SentinelOne Stock Craters After Earnings. It’s Cutting Jobs for AI. Barron's
Score: 42🌐 MovesMay 29, 2026https://www.barrons.com/articles/sentinelone-earnings-stock-price-4332e987 - Colombian army looks to outsmart guerrillas with drone warfare
SOGAMOSO (COLOMBIA) - A drone drops an explosive onto mountainous terrain below, shattering the silence of the Andes as Colombia's military enters an arms race with guerrillas already using the remote devices to wreak havoc.
Score: 42🌐 MovesMay 29, 2026https://www.bangkokpost.com/world/3263228/colombian-army-looks-to-outsmart-guerrillas-with-drone-warfare - Kaspersky warns of rising AI cyber threats in Kenya and East Africa
The company also advises families to maintain open discussions with children regarding their use of AI technologies and online safety practices
- Will AI be the end of jobs? The reality looks different in Canada
Canadians’ relatively slower approach to AI adoption and investment may prove a saving grace to its labour market
- Many autonomous agents doomed by governance failures
By next year, 40% of enterprises will have their autonomous AI efforts in part derailed by gaps in governance discovered only after production incidents, a recent report from Gartner predicts. The reason is that enterprises are treating AI agent governance as binary, either locked down or fully trusted, “and that is the root cause of failure,” said report author Shiva Varma , senior director analyst at Gartner. These failures will force enterprises to demote or decommission some agents. “Agents operate at different autonomy levels and across different trust boundaries . When the same controls are applied indiscriminately, organizations encounter two common failure modes: over-restriction of simple agents, which slows delivery and drives shadow development, or under-restriction of more autonomous agents, which increases operational, security and compliance risk,” he wrote. To combat this, Gartner recommends a multi-tiered governance approach, based on agents’ degree of autonomy. “Autonomy level and scope must be assessed independently,” Varma wrote. “Autonomy level defines an agent’s ability to act, while scope defines the breadth of data, systems and permissions it can access. Governance decisions should consider both dimensions, as risk increases with either expanded autonomy or expanded scope.” Gartner’s four level governance model, he said, only looks at autonomy level, since access controls scale separately. In the model, agents that have read-only access to defined data sources, and only display results to the requesting user, are designated Level 1, “Observe”. Governance of these agents, said Varma, should focus on baseline controls: scoped data access, user authentication, usage logging, and “basic functional and security testing.” Level 2 (“Advise”) agents, he said, also only have read-only access, but generate recommendations to users in activities such as email drafting, report or code generation, or decision support. But since their advice can affect human judgements, Level 1 constraints aren’t enough; they must be extended to include accuracy and hallucination testing, and domain-specific quality evaluation. In addition, user training needs to include information on the appropriate levels of reliance they should place on the results. At Level 3, “Act with Approval”, agents act with human approval, and perform tasks such as writing data, sending communications, or modifying configurations. It requires even stronger controls, building on Level 2 governance. “At this level, human review is effective only if it remains a meaningful control,” said Varma. “Without strong security testing, clear approval workflows with audit trails and agent‑specific incident response procedures, approvals can degrade under time pressure or approval fatigue, creating a false sense of safety while expanding the attack surface.” The toughest restrictions must be applied to fully autonomous agents at Level 4, Varma said. They can execute actions independently within defined guardrails, while humans only review exceptions and look at audit logs and aggregated outcomes. That means, in addition to the controls in levels 1-3, these agents should have comprehensive guardrail definitions, rollback capabilities for their actions, continuous monitoring, and a way to stop an agent’s operation if it violates its thresholds. In addition, there needs to be continuous red team testing, clear ownership and accountability for its actions, and business continuity procedures should the agent fail. Varma advised software engineering leaders to audit agents currently in use and match their governance level to their autonomy. Sanchit Vir Gogia , chief analyst at Greyhound Research, welcomed Gartner’s recommendations. “Applying one governance model to all agents is rather like applying the same control regime to a receptionist, a finance controller, a database administrator, a claims handler, and a procurement head because all of them use a laptop. It is tidy on paper. It is nonsense in practice.” To be effective, he said, governance models must recognize that the riskiest thing about an agent is not always what it says, but what it can do next. Valence Howden , advisory fellow at Info-Tech Research Group, agreed. “At [Level 4] the governance system must be adaptable and the organizations will need to move to more resilient anti-fragile adaptive models.” Gogia added, “The real governance problem is not model intelligence. It is delegated operational authority moving across trust boundaries faster than enterprises can instrument, constrain, or audit it. Governance is not a brake on AI adoption. It is the precondition for scaling it.” His advice to CIOs is blunt: “Do not scale agents faster than you can govern their authority. A small number of well-governed agents will create more enterprise value than a sprawling estate of clever, fragile, over-permissioned digital apprentices. The future of AI agents is not autonomy without restraint. It is autonomy inside well-designed boundaries.”
Score: 42🌐 MovesMay 29, 2026https://www.cio.com/article/4178628/many-autonomous-agents-doomed-by-governance-failures.html - Scaling safe enterprise AI with OpenAI governance frameworks
OpenAI’s latest governance frameworks offer enterprise leaders a structured blueprint for scaling safe and compliant AI deployments globally. The adoption of large language models has steadily progressed towards requiring sustainable, commercial-grade architecture. OpenAI has released its Frontier Governance Framework (FGF), documenting how the organisation addresses systemic risk assessment and mitigation. The framework maps directly to […] The post Scaling safe enterprise AI with OpenAI governance frameworks appeared first on AI News .
Score: 42🌐 MovesMay 29, 2026https://www.artificialintelligence-news.com/news/scaling-safe-enterprise-ai-openai-governance-frameworks/ - Amazon says it shut down a token leaderboard: 'Don't use AI just to use AI'
Amazon says it shut down a token leaderboard: 'Don't use AI just to use AI' Business Insider
Score: 42🌐 MovesMay 29, 2026https://www.businessinsider.com/amazon-ai-leaderboard-tokenmaxxing-2026-5 - Hands On: Acer's Predator Helios 18 Can Moonlight as an AI Warrior
Hands On: Acer's Predator Helios 18 Can Moonlight as an AI Warrior PCMag
Score: 42🌐 MovesMay 29, 2026https://www.pcmag.com/news/hands-on-acers-predator-helios-18-computex-2026 - Columbia-led team develops open-source framework to accelerate health AI research
Columbia-led team develops open-source framework to accelerate health AI research EurekAlert!
- AI ‘voice cloning’ scams are on the rise. Here’s how to protect yourself
AI ‘voice cloning’ scams are on the rise. Here’s how to protect yourself The Mercury News
Score: 41🌐 MovesMay 29, 2026https://www.mercurynews.com/2026/05/29/ai-voice-cloning-scams-are-on-the-rise-heres-how-to-protect-yourself/amp/ - Amazon Is Making an AI-Animated ‘Good Advice Cupcake’ TV Show. Its Original Creator Is Furious
Loryn Brantz created The Good Advice Cupcake for BuzzFeed years ago. The company licensed the character for a new Amazon series—made with AI—without her consent.