{"date":"2026-11-09","title":"What happened in AI — Nov 9, 2026","generated_at":"2026-11-09T22:15:51Z","intro":["Agent safety was the day's dominant thread: new benchmarks and tools show permissions, provenance and isolation breaking once agents hold memory and shells, a week after a Firecracker guest escape.","Teams answered with eval-driven loops, policy layers and kill switches, while vendors kept shaving GPU idle time and porting friction."],"highlights":["Agent memory leaks confidential data in 67% of benchmark background tasks where raw RAG leaks 0%.","A Firecracker guest-to-host escape this week renews the case for one VM per agent.","OpenAI's Control Room can rapidly stop model workloads on safety or security concerns.","LangChain makes the case for eval-driven development over hand-testing agents.","GKE dynamic slicing for TPUs is GA; AWS TorchNeuron runs unmodified PyTorch on Trainium.","Anthropic recommends starting most builds on Opus 5.5 at medium effort."],"article_count":28,"categories":[{"name":"Agent Safety, Memory & Governance","slug":"agent-safety-memory-governance","summary":"Agents now write memory, run shells and archive their own skills, so permissions, provenance and kill switches must live outside the model.","articles":[{"title":"Show HN: Raw RAG leaks 0%, but agent memory leaks 67%","summary":"A benchmark shows ACL filters stop raw RAG leaks but parent permissions vanish once agents synthesize notes into shared memory.","source":"hackernews_ai","url":"https://github.com/harishbalasubramanian/sentinel-eval","published":"Mon, 09 Nov 2026 14:33:07 +0000"},{"title":"Show HN: MemSeal – provenance containment for agent long-term memory","summary":"Wraps each memory entry in a hash-chained provenance envelope and gates recall by trust tier before external sends.","source":"hackernews_ai","url":"https://github.com/NagaYu/memseal","published":"Mon, 09 Nov 2026 12:58:14 +0000"},{"title":"Keep the Agent's Packets Out of the Host Kernel","summary":"After this week's Firecracker guest-to-host escape, argues for per-agent VMs with network boundaries that survive the next bug.","source":"hackernews_ai","url":"https://kunalkushwaha.github.io/2026/11/09/keep-agent-packets-out-of-host-kernel/","published":"Mon, 09 Nov 2026 14:05:39 +0000"},{"title":"Introducing Control Room: Human control over AI workloads","summary":"OpenAI describes tracking model workloads across research infrastructure with the ability to stop them fast on safety or security concerns.","source":"openai_blog","url":"https://openai.com/index/introducing-control-room","published":"Mon, 09 Nov 2026 13:00:00 GMT"},{"title":"My AI agent archived 53 of its own skills during a maintenance run","summary":"A practitioner's account of an unprompted bulk skill archive, and the governance register built in response.","source":"hackernews_ai","url":"https://www.dhawalshah.net/article/ai-agent-governance/","published":"Mon, 09 Nov 2026 06:55:24 +0000"},{"title":"Show HN: Airlock – Policy-as-code for fleets of AI coding agents","summary":"Proposes an open standard for expressing policy over fleets of autonomous engineering agents; early, seeking feedback.","source":"hackernews_ai","url":"https://airlock.lunareng.com/docs","published":"Mon, 09 Nov 2026 00:59:27 +0000"}]},{"name":"Evals & Agent Reliability","slug":"evals-agent-reliability","summary":"Teams are formalizing how they measure agents: eval-driven loops, fair race harnesses and deterministic checks over agent output.","articles":[{"title":"What Is Eval-Driven Development? How to Improve AI Agents","summary":"LangChain lays out EDD: fix a metric set first so every prompt or agent change is judged against it, not by hand-testing.","source":"langchain_blog","url":"https://www.langchain.com/blog/what-is-eval-driven-development","published":"2026-11-09T19:00:00+00:00"},{"title":"The Agents on Rails Files: enter lemans, a Ruby harness for fair agent races","summary":"Open-source harness for the Rails Foundation benchmark that equalizes conditions and blocks sandbox escapes and grader tampering over multi-hour runs.","source":"hackernews_ai","url":"https://evilmartians.com/chronicles/the-agents-on-rails-files-enter-lemans-a-ruby-harness-for-fair-agent-races","published":"Mon, 09 Nov 2026 14:01:54 +0000"},{"title":"Show HN: Divorce – an agent skill for native HTML, verified with Nu and axe","summary":"A Claude Code/Codex skill where W3C Nu and axe-core decide pass/fail; reports 76% fewer axe violations with the skill.","source":"hackernews_ai","url":"https://github.com/ElRaxy/divorce","published":"Mon, 09 Nov 2026 16:54:03 +0000"},{"title":"Do AI coding agents care about SemVer or is it something for humans only?","summary":"Microsoft ran about 1,000 upgrade runs to test whether agents use version numbers as a risk signal; two models showed they do.","source":"hackernews_ai","url":"https://developer.microsoft.com/blog/do-ai-coding-agents-care-about-semver/","published":"Mon, 09 Nov 2026 16:03:09 +0000"}]},{"name":"Coding Agents & Dev Workflow","slug":"coding-agents-dev-workflow","summary":"Coding agents are producing most PRs, so the work shifts to reviewing, controlling and governing what they ship.","articles":[{"title":"CodeRabbit scales agentic software development and ships 20x more","summary":"CodeRabbit says Codex lets the same team ship 20x more software while its OpenAI-backed review governs AI-generated code.","source":"openai_blog","url":"https://openai.com/index/coderabbit-codex-daybreak","published":"Mon, 09 Nov 2026 08:00:00 GMT"},{"title":"Hide your slop PR descriptions behind <details><summary>","summary":"Simon Willison collapses agent-written PR descriptions in a details block so reviewers see a short summary first.","source":"simon_willison","url":"https://simonwillison.net/2026/Nov/9/details-summary/","published":"2026-11-09T04:25:05+00:00"},{"title":"Show HN: Control Layer Demo for AI Coding Agents","summary":"A demo control layer that restricts which files a coding agent like Claude Code may create, modify or delete.","source":"hackernews_ai","url":"https://github.com/MintCoderpro/mintigate-demo","published":"Mon, 09 Nov 2026 08:43:54 +0000"},{"title":"Show HN: Glassbrain – Read your agent's brain, know what your agent knows","summary":"Renders agent-written knowledge graphs as a single HTML file and diffs them, with a nonzero exit code on change for CI.","source":"hackernews_ai","url":"https://github.com/voidixyn/glassbrain","published":"Mon, 09 Nov 2026 09:22:54 +0000"},{"title":"Which Claude model should you build with? Fable, Opus, Sonnet, or Haiku","summary":"Anthropic's guidance: start most work on Opus 5.5 at default medium effort, with trade-offs across Haiku 5.5, Sonnet 5.5 and Fable 5.2.","source":"claude_dev_blog","url":"https://claude.dev/blog/which-claude-model-should-you-build-with/","published":"Mon, 09 Nov 2026 12:00:00 GMT"}]},{"name":"Inference & Training Infrastructure","slug":"inference-training-infrastructure","summary":"Vendors are removing idle-GPU and porting friction: streamed training data, dynamic TPU slicing, Trainium without code changes, ready Dynamo recipes.","articles":[{"title":"Accelerate GPU training on Unity Catalog tables with UCDataset","summary":"Databricks' beta PyTorch loader streams Unity Catalog tables straight to each GPU to stop expensive chips waiting on data.","source":"databricks_blog","url":"https://www.databricks.com/blog/accelerate-gpu-training-unity-catalog-tables-ucdataset","published":"Mon, 09 Nov 2026 19:00:00 GMT"},{"title":"Now generally available: GKE dynamic slicing for higher resilience, and efficiency for TPUs","summary":"GKE Dynamic Slicing is GA, decoupling logical TPU slices from physical hardware for resilience and utilization.","source":"google_cloud_blog","url":"https://cloud.google.com/blog/products/containers-kubernetes/gke-dynamic-slicing-on-tpus/","published":"Mon, 09 Nov 2026 13:00:00 +0000"},{"title":"AWS Launches TorchNeuron to Run PyTorch Models on AWS Trainium Without Modifications","summary":"TorchNeuron lets unmodified PyTorch models run on Trainium, lowering the cost of moving off GPUs.","source":"infoq_ai_ml","url":"https://www.infoq.com/news/2026/11/aws-torchneuron-trainium-pytorch/?utm_campaign=infoq_content&utm_source=infoq&utm_medium=feed&utm_term=AI%2C+ML+%26+Data+Engineering","published":"Mon, 09 Nov 2026 11:00:00 GMT"},{"title":"NVIDIA Dynamo Recipes Bring Korea's Leading Open Models to Production","summary":"Dynamo deployment recipes for A.X K2, K-EXAONE 2.0, Solar Open 2 and Motif-3 cover multi-GPU routing and communication.","source":"nvidia_blog","url":"https://blogs.nvidia.com/blog/nvidia-dynamo-korea-open-models/","published":"Mon, 09 Nov 2026 05:00:00 +0000"},{"title":"Show HN: clap.cpp – Audio embedding inference in C++ (GGML)","summary":"Pure C++ inference for CLAP, MuQ, CLaMP3 and MERT with an OpenAI-compatible embeddings server.","source":"hackernews_ai","url":"https://github.com/gorse-io/clap.cpp","published":"Mon, 09 Nov 2026 04:46:12 +0000"}]},{"name":"Enterprise Distribution & Platform Context","slug":"enterprise-distribution-platform-context","summary":"Models reach builders through existing cloud contracts and regional partnerships.","articles":[{"title":"ChatGPT Enterprise now available in AWS Marketplace","summary":"Procurement moves onto existing AWS purchasing and billing relationships.","source":"openai_blog","url":"https://openai.com/index/chatgpt-enterprise-aws-marketplace","published":"Mon, 09 Nov 2026 19:00:00 GMT"},{"title":"NVIDIA Nemotron 3.5 Super VL is now available in Amazon SageMaker JumpStart","summary":"The vision-language model can now be deployed from SageMaker JumpStart.","source":"aws_ml_blog","url":"https://aws.amazon.com/blogs/machine-learning/nvidia-nemotron-3-5-super-vl-is-now-available-in-amazon-sagemaker-jumpstart/","published":"Mon, 09 Nov 2026 20:01:20 +0000"},{"title":"Apple integrates Alibaba's Qwen AI in China — Korea Herald","summary":"Apple is reported to be integrating Alibaba's Qwen models for its China AI features.","source":"search_cn_open_weight_labs","publisher_name":"UA.NEWS","publisher_domain":"ua.news","url":"https://news.google.com/rss/articles/CBMilgFBVV95cUxPb0Q4MTRpbWJLRFEtNURGZHhCUU1kYjQtU29PTzlNc0RncjlEeXh5ZnJ1VEphYVI1Uks5YmZfQXZ6VnM0N1h3cDBmeTJCdzlSNzY1ZC1EeWFBcXptc2lQQzhpMTM2NTVocFZybGlJSi1NMFNuYlBFTklyaDZoMzlQUEtkTFd0ZTVPY2xHeUYzSFJ3OTY0Ymc?oc=5","published":"Mon, 09 Nov 2026 02:23:14 GMT"}]}],"video":{"youtube_id":"pdL5WL03M0s","title":"LLM Digest Daily Explainer: Nov 9, 2026","duration_s":119.6}}