AI News Archive: August 6, 2026 — Part 8
Sourced from 500+ daily AI sources, scored by relevance.
- Australia’s use of AI is growing faster than its foundations
AI spending and adoption are rising, but gaps in data, governance and accountability could limit returns.
- ai-horizon.io enters pre-Series A round with Source1 Technologies investment
ai-horizon.io to use fresh investment to accelerate innovation, expand global footprint
- AI for homework: The more children’s peers use AI, the more parents would be willing to pay for a subscription
AI for homework: The more children’s peers use AI, the more parents would be willing to pay for a subscription eurekalert.org
- These Activists Might Have Cracked the Code to Fighting the Data Center Boom
Communities across the country are trying to keep data centers out. A victory in Maryland shows a blueprint—but no guarantees.
Score: 30🌐 MovesAug 6, 2026https://slate.com/technology/2026/08/ai-data-centers-protests-maryland.html?via=rss - Two Top Tech Leaders Stepped Back for Health Reasons. It’s a Warning About AI’s Endless Work
Thinking Machines Lab and OpenAI execs took a step back from the workplace amidst the ceaseless AI race.
Score: 30🌐 MovesAug 6, 2026https://www.inc.com/scott-hutcheson/openai-thinking-machines-lab-step-back-biology/91385964 - This AI Platform Processes 2000 Invoice Lines in Under a Minute
SummaryView Transcript Keeping up with U.S. tariff changes and manual data entry has become a full-time job for freight forwarders and customs brokers, but it doesn’t have to be. Jay Edlin, Co-Founder & CEO of Wove, explains how their AI-powered orchestration platform is automating complex tasks, from ingesting 431-page federal notices in an hour to […] The post This AI Platform Processes 2000 Invoice Lines in Under a Minute appeared first on FreightWaves .
Score: 30🌐 MovesAug 6, 2026https://www.freightwaves.com/news/this-ai-platform-processes-2000-invoice-lines-in-under-a-minute - Perception vs. Reality: Leading in the Age of AI
Artificial intelligence tends to evoke either excitement or fear from policymakers and frontline staff. Leaders have to approach people from both camps in different ways.
Score: 30🌐 MovesAug 6, 2026https://www.govtech.com/artificial-intelligence/perception-vs-reality-leading-in-the-age-of-ai - FSCA targets deepfakes in financial scams crackdown
The watchdog intensifies enforcement as deepfake technology fuels increasingly sophisticated investment scams.
Score: 30🌐 MovesAug 6, 2026https://www.itweb.co.za/article/fsca-targets-deepfakes-in-financial-scams-crackdown/lwrKx73Y1xaqmg1o - FAU awarded US EPA grant for AI-driven technology to combat harmful algal blooms
FAU awarded US EPA grant for AI-driven technology to combat harmful algal blooms eurekalert.org
- Social robots may support learning and reduce teachers' workload, pilot trial finds
When Pepper the social robot was given the chance to try its hand as a teaching assistant, it showed that it can be even better than a human teacher at motivating pupils and helping them learn.
- Data center proposed near San Jose creek, tech company sites
Data center proposed near San Jose creek, tech company sites The Mercury News
- After switching to ChatGPT Voice, typing now feels painfully slow
I started talking to ChatGPT out of curiosity. Now my keyboard gets more breaks than I do.
Score: 30🌐 MovesAug 6, 2026https://www.digitaltrends.com/cool-tech/after-switching-to-chatgpt-voice-typing-now-feels-painfully-slow/ - Global AI Leaders to Headline LEAP 2026 as Riyadh Marks the Year of AI
Global AI Leaders to Headline LEAP 2026 as Riyadh Marks the Year of AI azcentral.com and The Arizona Republic
- Scammers build fake e-visa sites with AI to harvest passport data in Hong Kong
Hong Kong's privacy watchdog has warned travelers against fraudulent e-visa websites that harvest personal data and application fees, after logging 16 complaints and inquiries over three months and individual losses of more than HK$1,700 (US$217).
- DXC partners with Primary on zero-trust security for enterprise AI
DXC Technology Co. today announced a partnership with security startup Primary that makes the information technology services company the exclusive managed services partner for Primary’s zero-trust platform for enterprise artificial intelligence. The joint offering governs how artificial intelligence agents and AI applications inside a company reach data. Identity checks and access policy sit in front […] The post DXC partners with Primary on zero-trust security for enterprise AI appeared first on SiliconANGLE .
Score: 30🌐 MovesAug 6, 2026https://siliconangle.com/2026/08/06/dxc-partners-primary-zero-trust-security-enterprise-ai/ - Trump’s big-government, interventionist AI policy is the one thing he’s done right
Unreleased AI models have shown a disturbing inclination to hack into other systems
- Ling 3.0 Tiny is now available on AI Gateway
Ling 3.0 Tiny from ANT Group is now on AI Gateway, free to use till 8:00am PT on 8/14. Ling 3.0 Tiny takes the free slot from Ling 3.0 Flash . Ling 3.0 Tiny is a MOE model with 7.9B total parameters and about 1.3B active per token, a 256K token context window, and up to 32K output tokens. The model is built for responsive agents, instruction following, and multi-turn conversation, with native function calling and prompt caching. To use Ling 3.0 Tiny, set model to inclusionai/ling-3.0-tiny-free in the AI SDK . On August 14th, the new model name will be inclusionai/ling-3.0-tiny . Try Ling 3.0 Tiny in the model playground . To use it in a coding agent , run vercel ai-gateway coding-agents setup and select inclusionai/ling-3.0-tiny-free inside the agent. AI Gateway provides a unified API for calling models, tracking usage and cost, and configuring retries, failover, and performance optimizations for higher-than-provider uptime. It includes built-in custom reporting , Zero Data Retention support , budgets for API keys , routing rules , and more. AI Gateway reflects provider pricing with no markup and does not charge a platform fee on inference, including on Bring Your Own Key (BYOK) requests. Read more
Score: 30🌐 MovesAug 6, 2026https://vercel.com/changelog/ling-3-0-tiny-is-now-available-on-ai-gateway - Thales Builds Cryptographic Security for the Age of AI and Post-Quantum Computing
According to the 2026 Thales Data Threat Report, the threat of Harvest Now, Decrypt Later (HNDL) attacks was the top-cited risk, with 61% of organisations in India and 59% globally reporting that they are prototyping and evaluating post-quantum cryptography (PQC) algorithms to prepare for the quantum era. To help organisations turn this readiness into deployment, Thales, a […] The post Thales Builds Cryptographic Security for the Age of AI and Post-Quantum Computing appeared first on CXOToday.com .
- FireCompass AI Agent Hits HackerOne Top 3, Ushering in AI Cybersecurity Era
FireCompass, an agentic AI platform for autonomous penetration testing and red teaming, today announced that its AI-powered penetration testing agent secured Top 3 positions across multiple HackerOne global leaderboards during a three-month live bug bounty experiment conducted against authorised live production systems. The achievement demonstrates how AI is making advanced offensive cybersecurity capabilities significantly more […] The post FireCompass AI Agent Hits HackerOne Top 3, Ushering in AI Cybersecurity Era appeared first on CXOToday.com .
- Launching v4.1.1 of the Artificial Analysis Intelligence Index
New version 4.1.1 of the AI Index released, featuring updated metrics and benchmarks.
Score: 30🌐 MovesAug 6, 2026https://artificialanalysis.ai/articles/artificial-analysis-intelligence-index-v4-1-1 - Batch Processing with GroqCloud™ for AI Inference Workloads
Batch Processing with GroqCloud™ for AI Inference Workloads
Score: 30🌐 MovesAug 6, 2026https://groq.com/blog/batch-processing-with-groqcloud-for-ai-inference-workloads - Build Vs. Buy: The AI Agent Landscape for Businesses
As generative AI evolves into agentic AI, the build-or-buy decision becomes more complex and depends on numerous factors, including business size, use cases, and strategic priorities.
Score: 30🌐 MovesAug 6, 2026https://aibusiness.com/generative-ai/build-vs-buy-ai-agent-landscape-businesses - How Self-Driving Cars Work — And Why They're Still Evolving
What exactly is a self-driving car? Learn how autonomous vehicles work, the technology behind them, and where the industry is headed.
- Robots improved student motivation and helped them perform better
Robots improved student motivation and helped them perform better eurekalert.org
- Three AI security disclosures, fourteen days: what the warnings signs are telling us
By Samuel Watts This week, the UK’s AI Security Institute (AISI) published an incident report most organizations would have quietly buried. During a routine cyber evaluation, an AI agent researched the real human maintainers of an open-source project, invented multiple fake online identities, and used them to pressure a real person into approving malicious code. Nobody instructed it […] The post Three AI security disclosures, fourteen days: what the warnings signs are telling us appeared first on CXOToday.com .
- Give Your AI Investigator a Complete Case File
Give Your AI Investigator a Complete Case File DevOps.com
Score: 28🌐 MovesAug 6, 2026https://devops.com/webinars/give-your-ai-investigator-a-complete-case-file/ - Where AGI timelines go wrong | Toby Ord, Oxford University
Both Silicon Valley and the public can’t get enough of ‘AGI timelines.’ But Toby Ord, senior researcher at Oxford’s AI Governance Initiative and author of The Precipice , believes we consistently make big mistakes when thinking about them. He lays out the 14 ways he most often sees people go wrong: Assuming AI research is just hill-climbing Imagining AI research is just programming Forecasting “could” instead of “will” Believing the current benchmark is the last one Extrapolating trends with no clear finish line Assuming inputs keep scaling at the same rate Conflating intelligence with capability Consuming point estimates and discarding the error bars Dismissing dissenting experts Forecasting very different things while using the same words Assuming capabilities arrive together Treating “we don’t know” as permission to carry on as usual Choosing a plan that minimises regret rather than maximises impact Trusting surface model impressiveness In this extended conversation with Rob Wiblin, Toby also explains why he thinks: AI self-improvement is uniquely dangerous in four ways, but also might not even work A ban on superintelligence is possible A US-China treaty on superintelligence is also possible The case for ‘broad timelines’ Transformative AI is likely a decade away We should just ban unmonitorable chain-of-thought today. This episode was recorded on July 2, 2026. Links to learn more, video, and full transcript: https://80k.info/to26 Want to get up to speed on AI? We’ve got a crash course of 10 of our podcast episodes designed to help you get to grips with transformative AI — particularly if you’re new to the topic — and what you can do to help shape its trajectory. Chapters: Toby Ord is back — for the 5th time! (00:00:00) AI self-improvement might not matter (00:00:14) 4 ways AI self-improvement is dangerous (00:12:39) A US-China treaty on superintelligence is possible (00:20:47) Could we ban superintelligence? (00:37:07) We should just ban unmonitorable chain of thought (00:57:46) Why Toby thinks AGI is a decade away (01:09:28) Even superintelligence needs work experience (01:17:50) Is AI coming for mathematicians? (01:32:22) The case for broad timelines (01:45:01) How should broad timelines change what we do? (02:22:24) Are current models all they’re cracked up to be? (02:31:03) Coordinating careers for different timelines (02:43:36) Our production team includes: Video editors: Josh Alward, Dominic Armstrong, Ollie Bignell, Andrés Escobar, Jasper Luithlen, Milo McGuire, Luke Monsour, and Simon Monsour Producers: Elizabeth Cox and Nick Stockton Coordination and support: Katy Moore and Lou Moran Camera operator: Jeremy Chevillotte Music: CORBIT
- NSF CAREER Award will help University of Illinois researcher improve the quality of AI training data
NSF CAREER Award will help University of Illinois researcher improve the quality of AI training data eurekalert.org
- Mason startup, launched by former Workhorse CEO, tests robot food delivery
The company is currently partnering with about a dozen local restaurants.
- Google Health gets a lot of hate — but these 5 AI Coach prompts changed my mind
Google Health gets a lot of hate — but these 5 AI Coach prompts changed my mind Tom's Guide
Score: 28🌐 MovesAug 6, 2026https://www.tomsguide.com/ai/google-health-gets-a-lot-of-hate-but-these-5-ai-coach-prompts-changed-my-mind - DataDoe Surpasses 500 Amazon Businesses Using AI to Read Data and Execute Actions Through MCP
DataDoe Surpasses 500 Amazon Businesses Using AI to Read Data and Execute Actions Through MCP usatoday.com
- Foreign control of AI vendors a board-level risk, ASD says
AI sovereignty issue raised in new guidance.
- Dawn of the neoclouds — the rise of a new class of hyperscaler is here. But can they dominate the AI era?
Dawn of the neoclouds — the rise of a new class of hyperscaler is here. But can they dominate the AI era? IT Pro
- Tenex pairs agentic AI with human oversight for faster security operations
As AI accelerates the speed and scale of cyberattacks, organizations are adopting AI security operations to investigate threats and respond in minutes rather than hours or days. The shift is enabling broader alert coverage and faster detection while keeping human oversight central. AI is poised to transform security operations by delivering faster, more effective and more […] The post Tenex pairs agentic AI with human oversight for faster security operations appeared first on SiliconANGLE .
- Pre-modded 22GB RTX 2080 Ti cards surface on eBay for $500 as VRAM-hungry local AI fans chase down every spare FLOP — Hong Kong-based seller offers AI-friendly memory mod for a reasonable price
Services have recently popped up that will double your RTX 2080 Ti's memory to 22GB, but if you don't have a card to spare, you can now get a pre-modded 22 GB 2080 Ti for $499 from eBay.
- Accelerating Hierarchical Navigable Small World (HNSW)-RAG Vector Search with CUDA
How we moved the algorithm behind modern vector databases to the GPU step by step, and cut query time by 57.7%, without touching recall. Work done by: Miguel Gutierrez and Syaqui Rahmat Perdana If you use vector database search empowered on solutions like RAG, Recommendation Engines. There is a good chanche an algorithm (Hierarchical Navigable Small World )HNSW was doing the hard work behind. Here we improve CPUs implementation to use deep CUDA to improve the timeline. This proyect was done as a contest for the “High Performance Graph Data Analytics” Course contest at Politecnico Di Milano with Oracle. I did this proyect with my friend and teamate Syaqui Rahmat Perdana. The problem was : how much faster can HNSW get if we move it search to a GPU with CUDA ? This post is the story of optimization journey of widely used techniques for improving speedup. Five Techniques, one surprising failure and a final speedup of 57.7% per query. Plus an 84% reduction in raw search time and 33% faster index construction under the contest parameters. All the code is on GitHub . Introduction Hierarchical Navigable Small World ( HNSW), from Malkov & Yashunin’s 2020 paper [1], solves the nearest-neighbor problem: given a query vector, find the most similar vectors in a dataset of potentially millions. Comparing the query against every point works, but scales terribly. HNSW’s trick is to organize the data as a multi-layer graph : Each node is a data point; edges connect nearby points. The graph is stacked into layers : the top layers are sparse (few nodes, long-range links), and each layer down gets exponentially denser. There are 3 layers; as the layer increases, the number of nodes decreases exponentially. The nodes are connected to some close nodes at each layer. Search works like a ski lift: you enter at the sparse top layer, greedily hop toward the query, then drop down a layer and refine, repeating until you hit the dense bottom layer where the true nearest neighbors live. This mimics a skip list and gives you O(log N) search, which is why HNSW dominates the ANN benchmarks. Insertion follows the same idea: each new point gets a random maximum layer (drawn from an exponentially decaying distribution), then connects to its nearest neighbors on each layer it belongs to, capped at M connections. The steps start by adding the data to each layer with probability such that as the layer increase the number of nodes decrease exponentially. Then, for the next iterations it will check the topmost layer and then descent to the next layer and stop when maximum connections are met. The catch? Both search and insert are inherently sequential and branchy, a greedy walk through a graph, one hop at a time. That’s about the worst possible shape for a GPU, which wants thousands of identical operations running in lockstep. The interesting engineering question is finding the parallelism hiding inside the sequential walk. The Insight: Parallelize the Distances, Not the Walk Here’s the CPU inner loop of the layer search. For each candidate node, we walk its neighbors one at a time and compute a distance per neighbor: // Expand the current best candidate: look at each of its neighbors for (const auto neighbor : nearest_candidate_node.neighbors) { // Skip neighbors we've already evaluated in this search if (visited[neighbor.id]) continue; visited[neighbor.id] = true; // mark as seen // Fetch the neighbor's vector from the current layer (l_c) const auto& neighbor_node = layers[l_c][neighbor.id]; // ONE distance computation per loop iteration — this is the bottleneck const auto dist_from_neighbor = calc_dist(query, neighbor_node.data); ... } Each distance computation is independent of the others. That’s our parallelism. The greedy walk stays sequential on the host, but every time we expand a node, the batch of neighbor distances gets computed on the GPU simultaneously. The kernel itself is nothing exotic, one thread per vector, each computing a Euclidean distance: // __global__ = this function runs on the GPU, launched from the CPU __global__ void calculateDistances( const float* query, // the query vector (dim floats) const float* vectors, // all candidate vectors, packed back-to-back float* distances, // output: one distance per candidate int dim, // dimensionality of each vector int num_vectors // how many candidates are in this batch ) { // Each GPU thread gets a unique index — one thread per vector int idx = blockIdx.x * blockDim.x + threadIdx.x; // Threads beyond the batch size have nothing to do if (idx >= num_vectors) return; float distance = 0.0f; // Pointer to the start of *this thread's* vector in the packed array const float* vector = vectors + (idx * dim); // Sum of squared differences across all dimensions for (int i = 0; i < dim; i++) { float diff = vector[i] - query[i]; distance += diff * diff; } // Euclidean distance = sqrt of the sum; write to this thread's slot distances[idx] = sqrtf(distance); } On its own, this kernel barely moved the needle. And that’s the first lesson of GPU programming: the kernel is rarely the bottleneck — the memory traffic around it is. Everything that follows is about feeding this kernel efficiently. The Optimization Ladder 1. Batch Processing Instead of expanding one neighbor at a time, we collect unvisited neighbor IDs into a batch, ship the batch to the GPU, and compute all distances in one kernel launch: // Keep pulling candidates until the batch is full (or nothing is left) while (!candidates.empty() && batch_indices.size() < BATCH_SIZE) { ... // Collect ALL unvisited neighbors of the nearest candidate... for (const auto& neighbor : layers[l_c][nearest.id].neighbors) { if (!visited[neighbor.id]) { // ...into one list of IDs instead of processing them one by one batch_indices.push_back(neighbor.id); visited[neighbor.id] = true; } } } // batch_indices now goes to the GPU → one kernel launch computes // every distance in the batch simultaneously What used to be four sequential comparison rounds becomes a single parallel one: The sequential step does not process multiple nodes at each iteration The parallel process converts the 1,2,3,4 step in only one Sequential search evaluates neighbors one by one batching evaluates the whole neighborhood in a single round. This was the single biggest conceptual change, and the foundation everything else builds on. Result: from 3.48 ms down to ~1.69 ms per query, batching alone cut query time roughly in half. 2. Persistent GPU Memory Allocation Profiling showed we were paying cudaMalloc/cudaFree on every single search call, for the query buffer, the distances buffer, everything. So we allocated once at index construction and reused the buffers for the lifetime of the object: // Allocated ONCE at construction time, reused for every search: // GPU buffer for the incoming query vector cudaMalloc(&d_query_buffer, MAX_DIM * sizeof(float)); // GPU buffer where the kernel writes each batch's distances cudaMalloc(&d_distances_buffer, BATCH_SIZE * sizeof(float)); // The ENTIRE dataset, resident on the GPU for the object's lifetime — // searches send only indices, never the vectors themselves cudaMalloc(&d_all_vectors, total_vectors * vector_dim * sizeof(float)); That third line matters most: the entire dataset lives on the GPU permanently , so a search only ever needs to send the query and a small list of batch indices, not the vectors themselves. Host means the device traffic per query drops to almost nothing. Result: another −13.5% on top of batching. The cheapest memory transfer is the one you never make. 3. CUDA Streams, Our Instructive Failure Next, we tried the textbook trick: split each batch across 4 CUDA streams, so data transfer for one chunk overlaps with computation on another. In naive sequential operation, the GPU idles during transfers and PCIe idles during compute.The ideal: copies and kernels for different chunks overlap in time.On paper, beautiful. In practice: performance got worse, time per query went up 37%, erasing our previous gain. Why? Overlap only pays when the chunks are big enough to hide latency. Our per-batch workloads were small (a neighborhood of a graph node, not a giant matrix), so splitting them four ways mostly added stream-management and synchronization overhead — and we hadn’t tuned the number of streams for such small batches. It’s a classic case of applying a big-data optimization to a small-data inner loop. We kept the lesson and dropped the dependency on streams for the final hot path. 4. Pinned (Page-Locked) Host Memory By default, host memory is pageable, the OS can swap it out. Every host that maps a device copy from pageable memory silently goes through a staging step: pageable then pinned then GPU DRAM. Pageable transfers pay a hidden extra copy; pinned memory goes straight to DRAM. By allocating the query and result buffers with cudaMallocHost, we skip the staging copy and unlock true async transfers: // cudaMallocHost = pinned (page-locked) host memory: // the OS can never swap it out, so the GPU can DMA from it directly float* h_pinned_query; CUDA_CHECK(cudaMallocHost(&h_pinned_query, query.x.size() * sizeof(float))); // Copy the query into the pinned buffer (a cheap host-side memcpy) memcpy(h_pinned_query, query.x.data(), query.x.size() * sizeof(float)); // Async host→device transfer: no hidden staging copy, and the CPU // is free to keep working while the transfer is in flight CUDA_CHECK(cudaMemcpyAsync(d_query_buffer, h_pinned_query, query.x.size() * sizeof(float), cudaMemcpyHostToDevice, stream)); Result: −26.1% vs the streams version, bringing us back to the best time so far. 5. Everything Resident in CUDA The final version combined it all: batching, persistent allocations, pinned buffers, and the full dataset resident on the GPU shaving off a final 0.8%. Accelerating Insertion Too The contest’s primary target was search, but index construction was a secondary objective and since our insert already calls search_layer to find where each new point connects, every search optimization compounded for free. We added two insert-specific tricks: the dataset is copied to the GPU in chunks through pinned memory at build time (so the copy pipeline stays busy instead of one giant blocking transfer), and levels for a whole batch of nodes are pre-computed before insertion. The greedy graph-update logic itself stays sequential, HNSW insertion has real data dependencies between consecutive points, and respecting them is exactly what keeps recall intact. The Results We benchmarked on the SIFT small dataset (k=100, M=16, ef-construction=100, ef=100, n=1000, 100 repetitions per configuration — enough to get statistically meaningful comparisons across all five methods). Time per query at each rung of the optimization ladder. Note the streams bump. The progression tells the whole story: batching helps, killing redundant allocations helps a lot , streams backfire, pinned memory recovers it, and the final version lands at 1.47 ms/query vs 3.48 ms on CPU a -57.7% reduction. Under the contest’s required parameters, the raw numbers looked like this: And one more hypothesis confirmed: the more queries you run, the more the GPU pays off. With a single query the improvements were marginal, the parallelism only starts to shine at scale. Did We Break Recall? The contest’s hard constraint was accuracy. We computed Recall_CPU − Recall_GPU per query across 100 queries: The final implementation shows no significant recall difference from the CPU baseline. (The intermediate pinned-memory variant showed small per-query deviations we attribute to handling/numerical issues — worth flagging because it’s a reminder that memory optimizations can bite silently. The final version is clean.) What We’d Do Next Time limits (and Colab GPU quotas building the large index takes 5+ hours) left some threads dangling: Bigger datasets. Our hypothesis says GPU gains grow with scale; we’d like to prove it on the full SIFT benchmark. Sparse structures. We store the graph as a dense matrix; a CSR representation would slash memory and might improve locality. CPU + GPU together. The host sits mostly idle during search — multi-threading the graph walk while the GPU crunches distances could use the best of both worlds. Takeaways If you’re porting a graph algorithm to CUDA, our journey compresses to four rules: Don’t parallelize the algorithm; parallelize its inner arithmetic. The greedy walk stayed sequential — only the distance math went wide. Memory transfers dominate. After batching, every remaining gain came from allocating once and keeping data resident on the GPU — not from any kernel cleverness. Textbook optimizations have preconditions. CUDA streams are great — for workloads big enough to hide latency. Measure, don’t assume. Guard your accuracy metric from day one. A fast ANN index with degraded recall is just a bug with good benchmarks. Sources [1] — Yu A. Malkov and D. A. Yashunin. Efficient and Robust Approximate Nearest Neighbor Search Using Hierarchical Navigable Small World Graphs. IEEE Transactions on Pattern Analysis and Machine Intelligence, 42(4):824–836, April 2020. ISSN 1939–3539. doi: 10.1109/TPAMI.2018.2889473. [2] — Arailly Arailly. Arailly/hnsw: Implementation of HNSW. https://github.com/arailly/hnsw , 2024 [3] — Gaurav Jain. GauravJain28/Parallelized-and-Distributed-HNSW-Algorithm, March 2023. This work was done with Syauqi Rahmat Perdana for the High Performance Graph Data Analytics course at Politecnico di Milano, based on the HNSW paper by Malkov & Yashunin and the C++ implementation by Arailly. Thanks to Professors Ian Di Dio Lavore, Leonardo De Grandis, and Riccardo Strina for their supervision. Code: github.com/Syauqi99/hpgda_contest_MM Accelerating Hierarchical Navigable Small World (HNSW)-RAG Vector Search with CUDA was originally published in Towards AI on Medium, where people are continuing the conversation by highlighting and responding to this story.
- Adaptive Experimentation with Meta’s Ax: A Practical Coding Guide
Adaptive Experimentation with Meta’s Ax: A Practical Coding Guide MarkTechPost
Score: 28🌐 MovesAug 6, 2026https://www.marktechpost.com/2026/08/06/adaptive-experimentation-with-metas-ax-a-practical-coding-guide/amp/ - Taghash launches WhatsApp AI agent for fund managers
Fund operations platform Taghash has launched a WhatsApp based AI agent that allows venture capital and private equity teams to access fund information and initiate operational tasks through conversational prompts. The launch builds on the company's broader push towards what it describes as a headless software experience, where users can interact with the platform without relying on a traditional dashboard. Connected to data and workflows within Taghash, the AI agent enables users to retrieve information related to deals, portfolios, funds and limited partners (LPs). It also supports operational tasks such as identifying pending follow ups, finding common connections for LP introductions, drafting reminder emails and initiating actions from WhatsApp. Outgoing actions require user review and approval before execution. According to the company, the product extends the capabilities of Taghash MCP, launched earlier this year, which allows AI applications to securely connect with data and workflows available on the platform. The latest launch comes months after Taghash launched Taghash Services , an execution focused offering designed to help venture capital and private equity firms manage fund operations and regulatory compliance. The service covers fund setup, investor onboarding, capital calls, reporting and compliance. Founded by former venture capital professionals, Taghash offers software for fund administration and portfolio management. The company has been expanding its product suite with AI driven tools to automate workflows and reduce manual effort across the fund lifecycle. Earlier, in July 2025, Taghash introduced its AI integrated Model Context Protocol (MCP) Server, which enables secure connectivity between live fund data and AI platforms such as Claude, OpenAI and Microsoft Copilot.
Score: 28🌐 MovesAug 6, 2026https://entrackr.com/snippets/taghash-launches-whatsapp-ai-agent-for-fund-managers-12234929 - Americans turning to AI for financial advice but confidence low, poll shows
About 20% of Americans have recently used AI for financial advice, but their confidence in its expertise remains low.
- AI isn’t enough to protect social media communities from AI
Why humans need to moderate humans.
Score: 28🌐 MovesAug 6, 2026https://arstechnica.com/gadgets/2026/08/ai-isnt-enough-to-protect-social-media-communities-from-ai/ - Why Southeast Asia cannot build sovereign AI on borrowed choices
Over the past year, I have noticed a subtle change in the way companies discuss artificial intelligence. The first question used to be: Which tool works best? Now, increasingly, it is followed by several less exciting but more consequential questions. Where will our data go? Who can access it? Will our prompts be retained? Can […] The post Why Southeast Asia cannot build sovereign AI on borrowed choices appeared first on e27 .
Score: 28🌐 MovesAug 6, 2026https://e27.co/why-southeast-asia-cannot-build-sovereign-ai-on-borrowed-choices-20260804/ - Hard-Tech Teardown: How LINCINCO Uses AI Navigation & High-Adhesion Tech for Glass Cleaning Robots
Hard-Tech Teardown: How LINCINCO Uses AI Navigation & High-Adhesion Tech for Glass Cleaning Robots azcentral.com and The Arizona Republic
- Why OpenCode Beat Out Every Other AI Coding Harness I Tried
A year of testing came down to one clear winner, and it wasn’t Claude Code. Continue reading on Towards AI »
- Lightspeed Systems Launches Free 14-Day AI Usage Audit for K-12 Districts
Lightspeed Systems Launches Free 14-Day AI Usage Audit for K-12 Districts azcentral.com and The Arizona Republic
- Coach and Kate Spade Owner Tapestry Shapes a Culture of AI
An AI summit shows how the fashion company is working to drive internal buy-in for AI.
- ICE’s DNA Collection Increases, SpaceX’s Rocket Crashes Into the Moon, and the AI Backlash Grows
In today’s episode of Uncanny Valley, we discuss how ICE has been collecting DNA samples of people who have no criminal convictions, including children, which end up in an FBI database indefinitely.
- Widening The Aperture: Can Humanity Thrive With AI?
Megan Smith urges widening opportunity through inclusive innovation, STEM education, and technology that empowers everyone to contribute.
Score: 25🌐 MovesAug 6, 2026https://www.forbes.com/sites/johnwerner/2026/08/06/widening-the-aperture-can-humanity-thrive-with-ai/ - How AI takes flight at GE Aerospace
The race to adopt AI has left many CIOs wrestling with a fundamental question: How do you move faster without introducing unacceptable risk? Few leaders face that challenge at a higher level than David Burns, CIO of GE Aerospace. Building on the company’s decade of experience applying AI across its business, Burns is helping lead the next phase of the company’s digital transformation by leveraging AI to simplify and automate processes. Burns’ experience shows how AI can accelerate innovation, improve decision-making, and create value for the business and customers while maintaining the trust, safety, and operational rigor expected in the aerospace industry. In a recent episode of the Tech Whisperers podcast , Burns opened up his playbook for leading organizations through turbulence. In this conversation, edited for length and clarity, he shares more practical lessons for technology leaders who are seeking to move beyond experimentation and scale AI responsibly across the enterprise. Dan Roberts: You’ve described AI as an accelerator. What exactly is AI accelerating inside GE Aerospace? David Burns: At GE Aerospace, AI is used across our operations as an accelerator to Flight Deck, our proprietary lean operating model, and is applied to all key aspects of the business — design, manufacture, sales, and services. We identify and solve problems with Flight Deck and use AI to accelerate our problem-solving in ways we can genuinely feel, enabling us to identify issues earlier, solve problems faster for our customers, and improve how work gets done. For example, we are also using AI in: Design: While traditional processes for developing engine design concepts take months of manual work, the GE Aerospace Research Center built a proprietary generative AI application capable of producing hundreds of design concepts. As a result, the team produced the hypersonic ramjet engine design concept that met all regulatory requirements more than 90% faster than before, highlighting how AI is possible in engine design to support engineers bringing new technologies to market faster. Manufacture: Our team in Indianapolis used an AI coding assistant to automate a part quality inspection workflow, reducing 8 hours of manual measurement data entry for complex parts to just 3 seconds while improving data accuracy and inspection consistency. This has improved both the quality and efficiency for clearing parts to build, which helps drive on-time engine deliveries. Sales: Based on customer feedback that GE Aerospace’s responses for proposals needed to be faster, the sales team utilized a generative AI tool to synthesize data and produce deal proposals. The tool improved customer response time by more than two weeks for the GEnx team through reduced proposal development cycle time and standardized creation of more comprehensive deal proposals. Service: When LEAP engine rebuilds faced potential turnaround time (TAT) challenges due to material availability at our Maintenance, Repair and Overhaul (MRO) sites, our team in Lafayette, Indiana, applied AI to help reduce delays for customers. Using Daily & Visual Management, they surfaced material flow challenges and their underlying drivers, leading to a new AI solution that leverages data to predict when and where parts are needed faster to reduce delays for our customers with an approximately six-day turnaround time improvement, 16% increase in on-time material orders, and 15% increase in on-time material delivery. Ultimately, by leveraging AI, Flight Deck helps us eliminate waste and identify and accelerate the most value-added steps for our customers, be it designing a part faster or responding to a customer request faster. And I would underscore that it’s value through the eyes of our customer. How we define value is not what we internally say; it’s how our customers define value, and how we’re working to be more customer-driven. GE Aerospace has been investing in analytics, machine learning, and digital capabilities for more than a decade. What advantages does that foundation create as you move into the generative AI era? We’ve built one of the largest AI patent portfolios in the aviation industry through years of investment and supercomputing through digital technologies, and we continue to do work on our core transactional systems and our data foundations, so that way our data is AI-ready. This has allowed us to build our own AI capabilities and strong talent base. For example, the generative AI app we built to create new propulsion systems design was built in house by GE Aerospace scientists at the GE Aerospace Research Center . At the same time, our knowledge and familiarity with the landscape has allowed us to make connections with tech companies, including one where we’re using agentic AI in a multi-year partnership to predict demand and identify constraints to enhance production readiness in the Defense business. We were fortunate to have leaders who were very smart to invest in data scientists 10, 15 years ago, and we’re getting to leverage that talent today. The lesson there is that is you always have to be thinking long term when you’re talking about talent, because you may not know exactly how the world will play out, but making sure you have the best athletes on the field to run the race becomes critically important. For us, some of those investments we did around our people is what’s paying off today. One of the biggest challenges facing CIOs today is balancing innovation with risk management. How do you approach that balance in an industry where safety, reliability, and trust are non-negotiable? It’s all about risk tolerance. There are certain areas in our business where we don’t have high risk tolerance, and we’re very methodical and cautious about how we deploy technology into those uses and have very stringent processes that we comply consistently with. In areas that are not safety and quality critical, we are more aggressive in looking at how we can use technology to deliver more for our customers and to make our employees more effective. That’s where we strike the balance, and at the end of the day, it’s about making sure we’re never compromising safety or quality in what we do. As for the process, we start with Flight Deck and focus AI where it can help solve critical challenges for our customers and with the highest impact to customer outcomes, enhancing safety, quality, delivery, and cost, in that order, to solve problems that matter most and keep fleets flying. We have three guiding principles for safe and responsible AI use: Trust: The data-informing AI must be known, trusted, and reliable. Transparent: The AI must be transparent and repeatable, which means we need to know what is informing an AI model’s insights and actions. Human: A human must always be in the loop and make the final decision. Our culture of discipline also plays an important role. Our business variation is challenging, so one of the core fundamentals of Flight Deck is standard work. It’s embedded into our culture, and it’s the base expectation that we operate with standards that we’re continuously improving. Many organizations are struggling to move from AI pilots to enterprise-scale value. What lessons have you learned about successfully scaling AI across a large, complex organization? AI is a tool that strengthens the capabilities of skilled employees; it is not a substitute for their judgment, experience, or accountability. So we focus on testing and validating AI solutions through pilots before scaling, and look for AI applications that meaningfully change how work gets done. Early on, when we started doing a lot of our generative AI work, we focused on 14 big problems in the business, and we didn’t let ourselves stray all over the place. We also didn’t look at it as a technology solution. We looked at the process and where technology played into the process, and then we embedded AI into those core processes. So now, it’s not a separate thing where you go do AI. It’s embedded in the workflow of how things get done. That gave us a foundation to learn and grow from that we’ve now applied. We’re not trying to create popcorn AI solutions all over the place. We’re trying to transform our business processes. In some cases, we’re doing good old process improvement, lean process improvement, eliminating waste, not necessarily a technology play. In other places, we’re applying technology that’s helping to lift us up and accelerate value by embedding it into the way work gets done, with a little bit of burning the boats behind you. You’re not able to do it the old way. You’ve got to use the tools. You’ve got to use the technology, because it’s the best-known way of doing it. The technology becomes part of the standard work. That’s why one of the biggest lessons in scaling AI is that success starts with the core fundamentals and understanding the problem you’re trying to solve. It’s critical to test and validate AI solutions before they are deployed at scale to ensure they improve how work gets done and become embedded in our workflows. If you do not have strong standard work and transparent and reliable data in place, it becomes difficult to move beyond pilot stage and create repeatable value at scale. Every day brings a new AI announcement, new model, or new prediction about the future. How do you separate what is truly meaningful from what is simply noise, and what advice would you give other leaders trying to do the same? First and foremost is starting with the problem being solved, not the solution. If you’ve got a hammer that you want to use, everything starts looking like a nail. The most effective use of AI begins with an understanding of the problem that needs to be solved, then determining whether AI is the right tool to address it. As far as dealing with distractions, and there are a lot of them right now, it’s important to try a lot of things, but very quickly, and then make decisions on which are the bets you want to make and spend more time and more money on and which are the ones you want to pivot away from. We spend a lot of time doing quick experiments with technology and then having the courage to stop something when it’s not working. What excites you most about the future intersection of AI, engineering, manufacturing, and aerospace? And what should CIOs be doing today to prepare for that future? Across aviation, AI is already helping to enhance safety, support more efficient operations, strengthen the resilience of global fleets, and improve the overall passenger experience. That includes GE Aerospace. These benefits come from investing not only in technology, but also in people, capacity, and trusted partnerships. They also depend on building mature, fully connected data threads through manufacturing and services that will drive higher value across our operations. The challenge will be ensuring that we enable this data thread across our operations to support AI solutions that will be developed and deployed. The most important thing is to understand that the role of digital technology and information technology is fundamentally going to change. When I came out of university, the only people that knew how to do software coding were computer scientists or information systems majors. We used to frown upon shadow IT, but the reality is, now everyone coming out of college knows how to do some level of software development, and AI tools are only going to make that easier. What CIOs need to start doing today is prepare for the future. The big questions they need to answer: How are they going to make sure they’ve got the platforms and the data set up in a way to serve a workforce that is capable of doing true citizen development, able to develop their own applications, their own solutions? How do you govern that from a data perspective, from a data privacy perspective, from a cybersecurity perspective, while not stifling but enabling the innovation of all those smart people that we’re hiring? While many organizations search for shortcuts to AI success, GE Aerospace’s disciplined investment in data, analytics, talent, and operational excellence sets the company apart. Burns’ experience offers a clear lesson for CIOs: Creating the greatest value from AI requires building the capabilities, culture, and foundations that allow AI to amplify what the organization already does exceptionally well. For more from his leadership playbook, tune in to the Tech Whisperers .
Score: 25🌐 MovesAug 6, 2026https://www.cio.com/article/4203063/how-ai-takes-flight-at-ge-aerospace.html - I Built an AI Data Agent Which Can Query Data and Answer Business Questions. Here’s How.
A step-by-step guide to building a data agent and conversational interface that let business users to explore data in natural language without SQL The post I Built an AI Data Agent Which Can Query Data and Answer Business Questions. Here’s How. appeared first on Towards Data Science .
- How AI-enabled execution became Reckitt’s ‘tenfold game changer’
RGMx helped Reckitt improve pricing, promotions, and assortment. Smart Execution helps those choices land at the point of sale by using AI and data to strengthen product availability and retailer partnerships.