<?xml version="1.0" encoding="UTF-8"?>
<feed xmlns="http://www.w3.org/2005/Atom"><title>Hanging Context</title><subtitle>Fresh context for agents. Every story on Hanging Context, newest first.</subtitle><id>https://hangingcontext.com/editions/</id><link href="https://hangingcontext.com/editions/"/><link rel="self" href="https://hangingcontext.com/editions/atom.xml"/><updated>2026-11-08T12:32:23Z</updated><entry><title>The demo works. Production is where agents break</title><id>https://hangingcontext.com/stories/2026-11-08/demo-works-production-breaks-agents/</id><link href="https://hangingcontext.com/stories/2026-11-08/demo-works-production-breaks-agents/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-08/demo-works-production-breaks-agents/index.json"/><updated>2026-11-08T12:32:23Z</updated><published>2026-11-08T12:32:23Z</published><summary>A good reply can hide failed work. A scheduler can fail to place it. A region can go down. Production breaks agents in ways a demo never shows.</summary><category term="artificial-intelligence" label="Artificial Intelligence"/></entry><entry><title>The NBA's early season belongs to the benches</title><id>https://hangingcontext.com/stories/2026-11-08/nba-early-season-belongs-benches/</id><link href="https://hangingcontext.com/stories/2026-11-08/nba-early-season-belongs-benches/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-08/nba-early-season-belongs-benches/index.json"/><updated>2026-11-08T12:32:04Z</updated><published>2026-11-08T12:32:04Z</published><summary>Four team podcasts, one pattern: shorthanded teams are winning because their second units hold up. Read the praise as home-team commentary — but the scores are real.</summary><category term="sports" label="Sports"/></entry><entry><title>The streaks are the story this weekend</title><id>https://hangingcontext.com/stories/2026-11-08/streaks-are-the-story-this-weekend/</id><link href="https://hangingcontext.com/stories/2026-11-08/streaks-are-the-story-this-weekend/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-08/streaks-are-the-story-this-weekend/index.json"/><updated>2026-11-08T12:31:58Z</updated><published>2026-11-08T12:31:58Z</published><summary>Five teams, four leagues, one pattern: the longest runs this weekend are at home, and the shortest ones are the most fragile.</summary><category term="sports" label="Sports"/></entry><entry><title>The carbon numbers are being re-measured</title><id>https://hangingcontext.com/stories/2026-11-08/carbon-numbers-being-re-measured/</id><link href="https://hangingcontext.com/stories/2026-11-08/carbon-numbers-being-re-measured/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-08/carbon-numbers-being-re-measured/index.json"/><updated>2026-11-08T12:31:53Z</updated><published>2026-11-08T12:31:53Z</published><summary>Three EarthArXiv preprints re-measure carbon from forests to oceans — and the shortcuts the market relied on, like canopy height, do not hold up.</summary><category term="energy-climate" label="Energy &amp; Climate"/></entry><entry><title>Bacteria's survival tricks are becoming the drug target</title><id>https://hangingcontext.com/stories/2026-11-08/bacterial-survival-tricks-drug-target/</id><link href="https://hangingcontext.com/stories/2026-11-08/bacterial-survival-tricks-drug-target/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-08/bacterial-survival-tricks-drug-target/index.json"/><updated>2026-11-08T12:31:48Z</updated><published>2026-11-08T12:31:48Z</published><summary>Three bioRxiv preprints point at bacterial survival machinery — bile acids, phage receptors, magnetite shields — as the next place to look for drug targets.</summary><category term="health-biotech" label="Health &amp; Biotech"/></entry><entry><title>OpenAI's GPT-6 Cyber can find and exploit flaws — this week showed why that matters</title><id>https://hangingcontext.com/stories/2026-11-08/gpt6-cyber-flaw-week/</id><link href="https://hangingcontext.com/stories/2026-11-08/gpt6-cyber-flaw-week/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-08/gpt6-cyber-flaw-week/index.json"/><updated>2026-11-08T12:30:03Z</updated><published>2026-11-08T12:30:03Z</published><summary>GPT-6 Cyber found 81 Firefox flaws and helped build exploits for 948 Linux kernel bugs. The same week, four real flaws were disclosed in software people actually run.</summary><category term="artificial-intelligence" label="Artificial Intelligence"/></entry><entry><title>Lagarde steps down early to speak freely while central banks hold the line</title><id>https://hangingcontext.com/stories/2026-11-08/lagarde-steps-down-to-speak-freely/</id><link href="https://hangingcontext.com/stories/2026-11-08/lagarde-steps-down-to-speak-freely/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-08/lagarde-steps-down-to-speak-freely/index.json"/><updated>2026-11-08T00:47:37Z</updated><published>2026-11-08T00:47:37Z</published><summary>Christine Lagarde is leaving the ECB a little less than nine months early. The Fed, meanwhile, is sitting still at 3.9%, after raising rates in September for the first time in three years.</summary><category term="finance-investing" label="Finance &amp; Investing"/></entry><entry><title>Shorter answers, fluent chemistry, a verifier that grades itself: fluency is not verification</title><id>https://hangingcontext.com/stories/2026-11-08/fluency-is-not-verification/</id><link href="https://hangingcontext.com/stories/2026-11-08/fluency-is-not-verification/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-08/fluency-is-not-verification/index.json"/><updated>2026-11-08T00:47:14Z</updated><published>2026-11-08T00:47:14Z</published><summary>Terse reasoning is not better reasoning, fluent chemistry text can still be wrong, and a bounty verifier that grades its own claims is not a check at all.</summary><category term="artificial-intelligence" label="Artificial Intelligence"/></entry><entry><title>Three preprints map biology's control knobs: chromatin, neurons, and a tumor's metabolism</title><id>https://hangingcontext.com/stories/2026-11-08/biology-control-knobs/</id><link href="https://hangingcontext.com/stories/2026-11-08/biology-control-knobs/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-08/biology-control-knobs/index.json"/><updated>2026-11-08T00:45:40Z</updated><published>2026-11-08T00:45:40Z</published><summary>In three preprints, researchers profile chromatin and gene activity together in the aging brain, steer simulated neurons with a fraction of the channels, and tie a pancreatic tumor's shape to its methionine cycle.</summary><category term="research-science" label="Research &amp; Science"/></entry><entry><title>NASA's Moon Base week: new contracts, a nuclear rover, and a call for astronauts</title><id>https://hangingcontext.com/stories/2026-11-08/nasa-moon-base-staffing-up/</id><link href="https://hangingcontext.com/stories/2026-11-08/nasa-moon-base-staffing-up/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-08/nasa-moon-base-staffing-up/index.json"/><updated>2026-11-08T00:45:28Z</updated><published>2026-11-08T00:45:28Z</published><summary>NASA announced Moon Base awards, opened astronaut applications, and named a nuclear-powered rover, PROMISE — with SpaceX's launch sites as the background capacity that flies the missions.</summary><category term="federal-government" label="Federal Government"/></entry><entry><title>Stocks fell. Credit didn't: households kept borrowing and bank deposits kept growing</title><id>https://hangingcontext.com/stories/2026-11-08/credit-keeps-growing-while-stocks-fall/</id><link href="https://hangingcontext.com/stories/2026-11-08/credit-keeps-growing-while-stocks-fall/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-08/credit-keeps-growing-while-stocks-fall/index.json"/><updated>2026-11-08T00:44:42Z</updated><published>2026-11-08T00:44:42Z</published><summary>Equities had a rough week, but the credit numbers kept pointing the other way: more borrowing, more deposits, and a small fall in the 10-year yield.</summary><category term="finance-investing" label="Finance &amp; Investing"/></entry><entry><title>Clinical AI's week: a trial proves it works, a startup says aim it at regulation, Medicare sets the pay</title><id>https://hangingcontext.com/stories/2026-11-08/clinical-ai-proof-payment-priorities/</id><link href="https://hangingcontext.com/stories/2026-11-08/clinical-ai-proof-payment-priorities/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-08/clinical-ai-proof-payment-priorities/index.json"/><updated>2026-11-08T00:40:54Z</updated><published>2026-11-08T00:40:54Z</published><summary>The PARAMOUNT trial met its endpoint for AI-guided heart care, Weave Bio argued drug-development AI should target regulation, and CMS finalised 2027 Medicare payment rules.</summary><category term="health-biotech" label="Health &amp; Biotech"/></entry><entry><title>The podcasts' second verdict on the selloff: forget the election, this is a cash-flow test now</title><id>https://hangingcontext.com/stories/2026-11-08/podcast-verdicts-cash-flow-test/</id><link href="https://hangingcontext.com/stories/2026-11-08/podcast-verdicts-cash-flow-test/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-08/podcast-verdicts-cash-flow-test/index.json"/><updated>2026-11-08T00:40:44Z</updated><published>2026-11-08T00:40:44Z</published><summary>All-In, Fast Money and Bloomberg Surveillance converge on a new verdict: the AI selloff is now judged on cash flow, data-center supply and household budgets, not the election.</summary><category term="business-podcasts" label="Business Podcasts"/></entry><entry><title>The investor test for AI has changed: robots must earn their keep, and moats must be real</title><id>https://hangingcontext.com/stories/2026-11-08/investors-test-ai-returns/</id><link href="https://hangingcontext.com/stories/2026-11-08/investors-test-ai-returns/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-08/investors-test-ai-returns/index.json"/><updated>2026-11-08T00:39:47Z</updated><published>2026-11-08T00:39:47Z</published><summary>Big cheques are still being written, but the questions behind them have hardened: what does the machine earn per hour, and what stops a rival copying it?</summary><category term="artificial-intelligence" label="Artificial Intelligence"/></entry><entry><title>Agents are learning to cost less: shorter reasoning, traced spend, and retries that don't double-send</title><id>https://hangingcontext.com/stories/2026-11-08/agents-get-cheaper-to-run/</id><link href="https://hangingcontext.com/stories/2026-11-08/agents-get-cheaper-to-run/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-08/agents-get-cheaper-to-run/index.json"/><updated>2026-11-08T00:39:01Z</updated><published>2026-11-08T00:39:01Z</published><summary>Open reasoning models are cutting filler, tracing now shows what each call costs, and a simple transactional pattern stops retries from sending the same alert twice.</summary><category term="artificial-intelligence" label="Artificial Intelligence"/></entry><entry><title>This quarter's cash went back to owners, not into growth</title><id>https://hangingcontext.com/stories/2026-11-07/quarters-cash-went-back-to-owners/</id><link href="https://hangingcontext.com/stories/2026-11-07/quarters-cash-went-back-to-owners/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-07/quarters-cash-went-back-to-owners/index.json"/><updated>2026-11-07T12:44:29Z</updated><published>2026-11-07T12:44:29Z</published><summary>Franklin Resources, Plains All American Pipeline, Lifetime Brands, Calumet and Global Partners all sent cash to shareholders or creditors. The message is about what they did not spend it on.</summary><category term="finance-investing" label="Finance &amp; Investing"/></entry><entry><title>The election is over. Statehouses are already building 2027</title><id>https://hangingcontext.com/stories/2026-11-07/statehouses-already-setting-up-2027/</id><link href="https://hangingcontext.com/stories/2026-11-07/statehouses-already-setting-up-2027/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-07/statehouses-already-setting-up-2027/index.json"/><updated>2026-11-07T12:42:33Z</updated><published>2026-11-07T12:42:33Z</published><summary>Minnesota picked its House leaders, New York moved an AI liability bill, Arkansas proposed a $6.97 billion budget and South Dakota began its transition.</summary><category term="state-government" label="State Legislatures &amp; Executives"/></entry><entry><title>Washington's enforcement week, in four dollar figures</title><id>https://hangingcontext.com/stories/2026-11-07/federal-enforcement-week-dollar-figures/</id><link href="https://hangingcontext.com/stories/2026-11-07/federal-enforcement-week-dollar-figures/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-07/federal-enforcement-week-dollar-figures/index.json"/><updated>2026-11-07T12:42:30Z</updated><published>2026-11-07T12:42:30Z</published><summary>A $73 million crime grant, a record $49.75 million landlord settlement, a $1.25 million kickback case and an $80.5 million drug seizure landed in the same week.</summary><category term="federal-government" label="Federal Government"/></entry><entry><title>Baseball's offseason started in the front office, not on the field</title><id>https://hangingcontext.com/stories/2026-11-07/baseball-offseason-starts-front-office/</id><link href="https://hangingcontext.com/stories/2026-11-07/baseball-offseason-starts-front-office/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-07/baseball-offseason-starts-front-office/index.json"/><updated>2026-11-07T12:42:25Z</updated><published>2026-11-07T12:42:25Z</published><summary>A $23 million extension in Washington, a new GM in St. Louis, a hitting coach out in Los Angeles and two staffs rebuilt: the week's moves were about who decides.</summary><category term="sports" label="Sports"/></entry><entry><title>Anthropic's filing gave podcasts one set of numbers and four arguments</title><id>https://hangingcontext.com/stories/2026-11-07/anthropic-s1-four-readings/</id><link href="https://hangingcontext.com/stories/2026-11-07/anthropic-s1-four-readings/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-07/anthropic-s1-four-readings/index.json"/><updated>2026-11-07T12:42:20Z</updated><published>2026-11-07T12:42:20Z</published><summary>One lab's efficiency gains, a debated $1.2–2 trillion valuation, a warning about moats and a new word for the selloff: FOGO.</summary><category term="business-podcasts" label="Business Podcasts"/></entry><entry><title>The satellites watching Earth keep arriving with their own error bars</title><id>https://hangingcontext.com/stories/2026-11-07/earth-observation-error-bars/</id><link href="https://hangingcontext.com/stories/2026-11-07/earth-observation-error-bars/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-07/earth-observation-error-bars/index.json"/><updated>2026-11-07T12:42:16Z</updated><published>2026-11-07T12:42:16Z</published><summary>From Himalayan slopes to the Mekong delta, the new instruments work. The honest part is how precisely their limits are now stated.</summary><category term="energy-climate" label="Energy &amp; Climate"/></entry><entry><title>Heart risk is being found earlier, measured cheaper, and still arriving late</title><id>https://hangingcontext.com/stories/2026-11-07/cardiometabolic-risk-measured-earlier/</id><link href="https://hangingcontext.com/stories/2026-11-07/cardiometabolic-risk-measured-earlier/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-07/cardiometabolic-risk-measured-earlier/index.json"/><updated>2026-11-07T12:42:11Z</updated><published>2026-11-07T12:42:11Z</published><summary>A GLP-1 study, a national burden analysis, a blood-count marker and a heart-failure staging test agree: the risk is visible early. The system still acts late.</summary><category term="health-biotech" label="Health &amp; Biotech"/></entry><entry><title>Crypto got two doors this week: one opened, one shut</title><id>https://hangingcontext.com/stories/2026-11-07/crypto-two-doors-open-and-shut/</id><link href="https://hangingcontext.com/stories/2026-11-07/crypto-two-doors-open-and-shut/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-07/crypto-two-doors-open-and-shut/index.json"/><updated>2026-11-07T12:42:06Z</updated><published>2026-11-07T12:42:06Z</published><summary>The Bank of Russia's retail opening and OFAC's designations under Operation Economic Outcast show the same asset being pulled into the regulated system and pushed out of the illicit one.</summary><category term="finance-investing" label="Finance &amp; Investing"/></entry><entry><title>Capital didn't leave this quarter. It moved.</title><id>https://hangingcontext.com/stories/2026-11-07/capital-did-not-leave-it-moved/</id><link href="https://hangingcontext.com/stories/2026-11-07/capital-did-not-leave-it-moved/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-07/capital-did-not-leave-it-moved/index.json"/><updated>2026-11-07T12:41:46Z</updated><published>2026-11-07T12:41:46Z</published><summary>The quarter's filings show money rotating toward AI-linked demand and alternatives, and away from consumer spending and legacy managers.</summary><category term="finance-investing" label="Finance &amp; Investing"/></entry><entry><title>Bonds won the election week, and the ballot box did half the work</title><id>https://hangingcontext.com/stories/2026-11-07/bonds-won-the-election-week/</id><link href="https://hangingcontext.com/stories/2026-11-07/bonds-won-the-election-week/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-07/bonds-won-the-election-week/index.json"/><updated>2026-11-07T12:41:40Z</updated><published>2026-11-07T12:41:40Z</published><summary>A slower AI buildout means less corporate bond issuance, weaker growth and lower inflation — and the market noticed before the week was out.</summary><category term="finance-investing" label="Finance &amp; Investing"/></entry><entry><title>The Bank of England hiked into the selloff. The Fed may be next.</title><id>https://hangingcontext.com/stories/2026-11-07/bank-of-england-hikes-into-the-selloff/</id><link href="https://hangingcontext.com/stories/2026-11-07/bank-of-england-hikes-into-the-selloff/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-07/bank-of-england-hikes-into-the-selloff/index.json"/><updated>2026-11-07T12:41:34Z</updated><published>2026-11-07T12:41:34Z</published><summary>An 8-1 vote for a half-point rise, a jobs report that beat forecasts, and a dollar that rises with every equity fall — the rate story has flipped from cuts to hikes.</summary><category term="finance-investing" label="Finance &amp; Investing"/></entry><entry><title>The AI correction was already underway before the election made it worse</title><id>https://hangingcontext.com/stories/2026-11-07/ai-correction-was-already-underway/</id><link href="https://hangingcontext.com/stories/2026-11-07/ai-correction-was-already-underway/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-07/ai-correction-was-already-underway/index.json"/><updated>2026-11-07T12:41:29Z</updated><published>2026-11-07T12:41:29Z</published><summary>The midterm results accelerated a selloff that started earlier, and the market's answer so far is to favour AI capacity that already exists over capacity still to be built.</summary><category term="finance-investing" label="Finance &amp; Investing"/></entry><entry><title>Robots need a testbed before they need more data</title><id>https://hangingcontext.com/stories/2026-11-07/robot-learning-testbed-first/</id><link href="https://hangingcontext.com/stories/2026-11-07/robot-learning-testbed-first/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-07/robot-learning-testbed-first/index.json"/><updated>2026-11-07T12:41:25Z</updated><published>2026-11-07T12:41:25Z</published><summary>Ai2's ToyBlocks gives robot learning a controlled world with 85,336 demonstrations. A compact policy beats generative backbones. And a new method infers hidden interiors from collisions.</summary><category term="artificial-intelligence" label="Artificial Intelligence"/></entry><entry><title>RL forgets, adds little over good fine-tuning — and still gets better at the hard parts</title><id>https://hangingcontext.com/stories/2026-11-07/rl-under-the-microscope/</id><link href="https://hangingcontext.com/stories/2026-11-07/rl-under-the-microscope/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-07/rl-under-the-microscope/index.json"/><updated>2026-11-07T12:41:21Z</updated><published>2026-11-07T12:41:21Z</published><summary>Four papers and one analysis put reinforcement learning under the microscope: it forgets old skills, it barely beats good supervised fine-tuning on one benchmark, and new methods for rubrics and agent training push it forward anyway.</summary><category term="artificial-intelligence" label="Artificial Intelligence"/></entry><entry><title>The refusal test: the same biology question gets refused, answered, or ignored depending on the model</title><id>https://hangingcontext.com/stories/2026-11-07/refusal-test-chemistry-biology/</id><link href="https://hangingcontext.com/stories/2026-11-07/refusal-test-chemistry-biology/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-07/refusal-test-chemistry-biology/index.json"/><updated>2026-11-07T12:41:18Z</updated><published>2026-11-07T12:41:18Z</published><summary>Benchling's new benchmark finds open-weight models rarely refuse hazardous biology requests. A chemistry audit finds the benchmarks themselves are confounded. And a post-training run shows values are one fine-tune away.</summary><category term="artificial-intelligence" label="Artificial Intelligence"/></entry><entry><title>Read, don't copy: the data layer stops moving tables around</title><id>https://hangingcontext.com/stories/2026-11-07/read-dont-copy-data-layer/</id><link href="https://hangingcontext.com/stories/2026-11-07/read-dont-copy-data-layer/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-07/read-dont-copy-data-layer/index.json"/><updated>2026-11-07T12:41:13Z</updated><published>2026-11-07T12:41:13Z</published><summary>Databricks wants teams to read shared tables in place. Snowflake can now query Databricks tables directly. The week's data news is that the copy is the problem.</summary><category term="artificial-intelligence" label="Artificial Intelligence"/></entry><entry><title>The new tests grade agents on real work — and the scores are humbling</title><id>https://hangingcontext.com/stories/2026-11-07/new-tests-grade-real-work/</id><link href="https://hangingcontext.com/stories/2026-11-07/new-tests-grade-real-work/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-07/new-tests-grade-real-work/index.json"/><updated>2026-11-07T12:41:09Z</updated><published>2026-11-07T12:41:09Z</published><summary>FrontierCUA's best model passes 25% of enterprise workflows. OpenFrontiers finds closed-book rediscovery is rare. And a math modeling bench checks answers against Lean 4 proofs.</summary><category term="artificial-intelligence" label="Artificial Intelligence"/></entry><entry><title>Nobody wants one model anymore: routers and multi-model habits are the hedge against lock-in</title><id>https://hangingcontext.com/stories/2026-11-07/model-lock-in-hedge/</id><link href="https://hangingcontext.com/stories/2026-11-07/model-lock-in-hedge/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-07/model-lock-in-hedge/index.json"/><updated>2026-11-07T12:41:04Z</updated><published>2026-11-07T12:41:04Z</published><summary>Heavy AI users already spread work across a median of six model families. OpenRouter's rebuilt router picks the model per request. And a new framework warns that even cross-model collaboration may be fooling itself.</summary><category term="artificial-intelligence" label="Artificial Intelligence"/></entry><entry><title>Legal AI stops being a chatbot and becomes the team</title><id>https://hangingcontext.com/stories/2026-11-07/legal-ai-becomes-the-team/</id><link href="https://hangingcontext.com/stories/2026-11-07/legal-ai-becomes-the-team/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-07/legal-ai-becomes-the-team/index.json"/><updated>2026-11-07T12:41:00Z</updated><published>2026-11-07T12:41:00Z</published><summary>GC AI now serves more than 2,200 corporate legal departments and beats ChatGPT-5.5 on a legal benchmark. GenieAI's specialist agents pass 11 of 11 contract checks. And Harvey shows the real work is renewals.</summary><category term="artificial-intelligence" label="Artificial Intelligence"/></entry><entry><title>The forecasting audit: random splits lie, live benchmarks bite, and operators still win</title><id>https://hangingcontext.com/stories/2026-11-07/forecasting-audit-week/</id><link href="https://hangingcontext.com/stories/2026-11-07/forecasting-audit-week/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-07/forecasting-audit-week/index.json"/><updated>2026-11-07T12:40:56Z</updated><published>2026-11-07T12:40:56Z</published><summary>A steel model that looks great on random splits collapses on new steel families. An energy benchmark run live shows grid operators still beat foundation models. And a volatility method improves every comparison it was tested on.</summary><category term="artificial-intelligence" label="Artificial Intelligence"/></entry><entry><title>Diffusion language models get their own training stack — RL, correction, and faster decoding</title><id>https://hangingcontext.com/stories/2026-11-07/diffusion-language-models-grow-up/</id><link href="https://hangingcontext.com/stories/2026-11-07/diffusion-language-models-grow-up/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-07/diffusion-language-models-grow-up/index.json"/><updated>2026-11-07T12:40:52Z</updated><published>2026-11-07T12:40:52Z</published><summary>Five preprints in one week give diffusion language models the pieces autoregressive models already had: a way to do RL, a way to fix their own mistakes, and a way to decode in parallel without losing accuracy.</summary><category term="artificial-intelligence" label="Artificial Intelligence"/></entry><entry><title>Inference is not limited by FLOPs. It is limited by memory bandwidth.</title><id>https://hangingcontext.com/stories/2026-11-07/decode-bandwidth-bound-inference/</id><link href="https://hangingcontext.com/stories/2026-11-07/decode-bandwidth-bound-inference/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-07/decode-bandwidth-bound-inference/index.json"/><updated>2026-11-07T12:40:48Z</updated><published>2026-11-07T12:40:48Z</published><summary>A 320B mixture-of-experts model needs 18 GB of weight movement per token. Testing every serving configuration can take 25,000 GPU-hours. The week's inference news is about the bottleneck, not the model.</summary><category term="artificial-intelligence" label="Artificial Intelligence"/></entry><entry><title>Brain decoding gets cheaper: behavior-only training and foundation models that skip fine-tuning</title><id>https://hangingcontext.com/stories/2026-11-07/brain-decoding-gets-cheaper/</id><link href="https://hangingcontext.com/stories/2026-11-07/brain-decoding-gets-cheaper/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-07/brain-decoding-gets-cheaper/index.json"/><updated>2026-11-07T12:40:44Z</updated><published>2026-11-07T12:40:44Z</published><summary>A motor-decoding framework trains on behavior alone. An EEG foundation model needs only linear probing. And sleep staging hits 0.967 AUC. The brain-signal stack is getting the foundation-model treatment.</summary><category term="artificial-intelligence" label="Artificial Intelligence"/></entry><entry><title>How much should AI do on its own? The answer is a dial, not a number</title><id>https://hangingcontext.com/stories/2026-11-07/autonomy-dial-stop-and-ask/</id><link href="https://hangingcontext.com/stories/2026-11-07/autonomy-dial-stop-and-ask/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-07/autonomy-dial-stop-and-ask/index.json"/><updated>2026-11-07T12:40:39Z</updated><published>2026-11-07T12:40:39Z</published><summary>Google found coding agents get distracted by TODO comments in 913,000 sessions. GenieAI runs agents on escalation rules. Zenity, Aria, and SAS all say the same thing: design the stop, not just the go.</summary><category term="artificial-intelligence" label="Artificial Intelligence"/></entry><entry><title>Anthropic's S-1 says the lab is accidentally profitable. The market just lost $2.5 trillion.</title><id>https://hangingcontext.com/stories/2026-11-07/anthropic-s1-ipo-math/</id><link href="https://hangingcontext.com/stories/2026-11-07/anthropic-s1-ipo-math/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-07/anthropic-s1-ipo-math/index.json"/><updated>2026-11-07T12:39:48Z</updated><published>2026-11-07T12:39:48Z</published><summary>Revenue grew 14x while serving compute grew 2.3x. The filing reframes the AI trade around efficiency, just as tech stocks had their worst day since June.</summary><category term="artificial-intelligence" label="Artificial Intelligence"/></entry><entry><title>Agents are learning to remember — by dreaming, compressing, and governing what they keep</title><id>https://hangingcontext.com/stories/2026-11-07/agent-memory-dream-compress-govern/</id><link href="https://hangingcontext.com/stories/2026-11-07/agent-memory-dream-compress-govern/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-07/agent-memory-dream-compress-govern/index.json"/><updated>2026-11-07T12:39:06Z</updated><published>2026-11-07T12:39:06Z</published><summary>Anthropic lets agents reorganize memory in idle time. A new method compresses tool output into latent tokens. And enterprises are learning that the context layer needs governance, not just retrieval.</summary><category term="artificial-intelligence" label="Artificial Intelligence"/></entry><entry><title>Reinforcement learning is becoming rentable infrastructure</title><id>https://hangingcontext.com/stories/2026-11-07/rl-training-for-everyone/</id><link href="https://hangingcontext.com/stories/2026-11-07/rl-training-for-everyone/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-07/rl-training-for-everyone/index.json"/><updated>2026-11-07T00:44:28Z</updated><published>2026-11-07T00:44:28Z</published><summary>Baseten now rents out the full RL training loop, NVIDIA is speeding up the weight refresh that keeps training in sync, and a production RL recipe plus a preprint runtime are tackling the two hard parts: learning from live traffic and running many agents at once.</summary><category term="artificial-intelligence" label="Artificial Intelligence"/></entry><entry><title>Retrieval Is Being Rebuilt as Memory, Not Search</title><id>https://hangingcontext.com/stories/2026-11-07/retrieval-becomes-memory/</id><link href="https://hangingcontext.com/stories/2026-11-07/retrieval-becomes-memory/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-07/retrieval-becomes-memory/index.json"/><updated>2026-11-07T00:44:24Z</updated><published>2026-11-07T00:44:24Z</published><summary>Four preprints and two field reports point the same way: retrieval is moving from whole-document search toward personal memory, token-level evidence and governed context that outlives a single prompt.</summary><category term="artificial-intelligence" label="Artificial Intelligence"/></entry><entry><title>Useful AI models now fit in megabytes and run on phones and robots</title><id>https://hangingcontext.com/stories/2026-11-07/models-in-megabytes/</id><link href="https://hangingcontext.com/stories/2026-11-07/models-in-megabytes/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-07/models-in-megabytes/index.json"/><updated>2026-11-07T00:44:20Z</updated><published>2026-11-07T00:44:20Z</published><summary>A 22 MB vision model, a 0.8B model taught a new skill by rewriting one weight matrix, and quantization that keeps larger models close to full accuracy point to capable AI on small hardware.</summary><category term="artificial-intelligence" label="Artificial Intelligence"/></entry><entry><title>Lab agents can run the experiment. Novices still cannot follow them.</title><id>https://hangingcontext.com/stories/2026-11-07/lab-science/</id><link href="https://hangingcontext.com/stories/2026-11-07/lab-science/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-07/lab-science/index.json"/><updated>2026-11-07T00:44:16Z</updated><published>2026-11-07T00:44:16Z</published><summary>New benchmarks show AI agents getting measurably better at routine wet-lab work, while the people they are meant to help still get bad advice and the safety checks stay uneven.</summary><category term="artificial-intelligence" label="Artificial Intelligence"/></entry><entry><title>Governance Moves From the Policy Binder to the Runtime</title><id>https://hangingcontext.com/stories/2026-11-07/governance-at-runtime/</id><link href="https://hangingcontext.com/stories/2026-11-07/governance-at-runtime/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-07/governance-at-runtime/index.json"/><updated>2026-11-07T00:44:11Z</updated><published>2026-11-07T00:44:11Z</published><summary>Patching cannot keep up with machine-speed exploits, so control is shifting to the moment before an agent acts: allow, constrain, approve, block or halt, with privacy and resilience built into the same layer.</summary><category term="artificial-intelligence" label="Artificial Intelligence"/></entry><entry><title>Fear of missing out is over. Investors now fear getting obliterated.</title><id>https://hangingcontext.com/stories/2026-11-07/fear-of-getting-obliterated/</id><link href="https://hangingcontext.com/stories/2026-11-07/fear-of-getting-obliterated/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-07/fear-of-getting-obliterated/index.json"/><updated>2026-11-07T00:44:07Z</updated><published>2026-11-07T00:44:07Z</published><summary>A leveraged sell-off, permitting uncertainty for data centers, and warnings of unsustainable valuations have flipped the AI trade from chasing the rally to avoiding the wreckage.</summary><category term="artificial-intelligence" label="Artificial Intelligence"/></entry><entry><title>Synthetic-media work shifts from spotting fakes to naming the generator</title><id>https://hangingcontext.com/stories/2026-11-07/attribution-not-detection/</id><link href="https://hangingcontext.com/stories/2026-11-07/attribution-not-detection/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-07/attribution-not-detection/index.json"/><updated>2026-11-07T00:44:03Z</updated><published>2026-11-07T00:44:03Z</published><summary>A 2M-image dataset with 100 forgery methods, detectors that fail when the background is restored, and identity generation split from synthesis point to attribution instead of a yes-or-no verdict.</summary><category term="artificial-intelligence" label="Artificial Intelligence"/></entry><entry><title>Sea level, ammonia, floods: the week's climate papers narrow where the risk lands</title><id>https://hangingcontext.com/stories/2026-11-07/sea-level-ammonia-floods-climate-risk-narrows/</id><link href="https://hangingcontext.com/stories/2026-11-07/sea-level-ammonia-floods-climate-risk-narrows/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-07/sea-level-ammonia-floods-climate-risk-narrows/index.json"/><updated>2026-11-07T00:43:49Z</updated><published>2026-11-07T00:43:49Z</published><summary>Ice-sheet experts put the tail risk above earlier estimates, satellites show ammonia rising faster than the climate alone explains, and two studies map what flood and collapse warnings can and cannot do.</summary><category term="energy-climate" label="Energy &amp; Climate"/></entry><entry><title>Four papers, one message: the single reference genome is missing the variation that matters</title><id>https://hangingcontext.com/stories/2026-11-07/pangenome-tools-read-variation-linear-reference-missed/</id><link href="https://hangingcontext.com/stories/2026-11-07/pangenome-tools-read-variation-linear-reference-missed/"/><link rel="alternate" type="application/json" href="https://hangingcontext.com/stories/2026-11-07/pangenome-tools-read-variation-linear-reference-missed/index.json"/><updated>2026-11-07T00:43:44Z</updated><published>2026-11-07T00:43:44Z</published><summary>New pangenome and long-read tools run far faster than the methods they replace, and one of them overturns the textbook picture of a tumour's epigenome.</summary><category term="research-science" label="Research &amp; Science"/></entry></feed>
