All AI updates

How backlash and red tape continue to slow data centers post-Blue Wave

TechCrunch's Equity podcast episode features hosts Rebecca Bellan and Tim De Chant with Brent Nelson of Ascend Analytics discussing how local opposition, permitting delays and power constraints are slowing data centers, and why Nelson believes the announced project pipeline is wildly inflated. The episode also covers what happens to AI infrastructure if the bubble bursts, why Democrats should pursue permitting reform now, why natural gas isn't the reliable fallback developers assume, and pairing battery storage with a curtailable grid connection as an alternative.

Meta is losing its longtime head of infrastructure

Santosh Janardhan, Meta's longtime head of infrastructure and co-head of engineering, is stepping down and will be replaced by Dave Brown, who joined as vice president of infrastructure earlier this year after 19 years at Amazon. Janardhan, who joined Meta in 2009, will stay on for a transition and move into an advisory role on Meta Compute, with his departure expected next year. Bloomberg first reported the move, citing an internal memo in which Janardhan cited the need to "pass the baton."

Why the data center you dread might never even be built

TechCrunch's Equity podcast discusses how the data center boom faces local opposition, permitting delays and a scramble for electricity, but notes that many announced projects may never be built at all. Hosts Rebecca Bellan and Tim De Chant are joined by Brent Nelson of Ascend Analytics to separate realistic data center projects from speculative ones amid billions of dollars in AI infrastructure investment.

I think differently, and I think AI might help

The author reflects on a Fortune piece about Palantir CEO Alex Karp's view that neurodivergent people may be well placed for the AI era, agreeing partly from personal experience with ADD-like traits. They argue AI handles the mechanical parts of work—drafting, summarizing, organizing rough notes—while the valuable human skills become challenging assumptions, spotting odd patterns and judging which questions are worth asking. They also caution that better tools won't fix rigid workplaces, and that neurodivergent people shouldn't have to be exceptional to earn a place at work.

Gemini Decision 1 is Free on Google AI Studio

A Hacker News submission notes that Gemini Decision 1 is available for free on Google AI Studio, linking to the AI Studio Decisions API documentation. The post had 6 points and no comments at the time of collection.

AI as a Lifestyle Choice

Graham Dumpleton, creator of mod_wsgi, describes using AI coding tools during his sabbatical to build Muru, a new Python web application hosting system, working seven days a week because he finds the process engaging and fun. He explains that AI removes the typing he finds a chore while leaving him the ideas and judgment, compares incremental improvement to the Japanese craft of making a dorodango, and credits the Claude for Open Source Program with freeing him from cost worries. He invites readers to review Muru's documentation and submit ideas via the GitHub discussions ideas section about what they would want in a Python hosting system.

AI Models Can Tell When Researchers Plant Concepts in Their Neural Networks

New research suggests large language models can sometimes detect when researchers artificially inject concept-related activity patterns into their neural networks, with Alibaba's Qwen3-235B flagging injections up to 53.9 percent of the time and Meta's Llama 3.1 405B up to 31.7 percent, while neither model claimed an injection in unaltered trials. However, the ability to name the injected concept appears distinct from detection and may stem from a side effect that makes models talk about the concept regardless of the question. Skeptics at New York University found models reported "injected thoughts" more often after a prompt than a genuine injection, raising doubts about whether the findings amount to introspection.

Is AI going to arb away all market inefficiencies?

The author describes reading books to challenge his belief that AI trading bots will eventually eliminate all market edge, noting he has long disagreed with the Efficient Market Hypothesis, like Edward Thorp. He quotes a passage from "The Migration of Scarcity" arguing that a market forecast is part of the market rather than a description of an independent state, so forecasting becomes coordination.

An open specification for agentic AI security evaluation and testing, from Cisco

Cisco's Advanced Security Initiatives Group released Foundry, an organization-neutral specification distilling its internal agentic security evaluation design into eight core agent roles, five extension roles, a finding lifecycle, coordination and governance requirements, and roughly 130 functional requirements with rationale. The repository contains no code by design: it is written to be run through a spec-driven workflow such as spec-kit, where organization-specific decisions marked [NEEDS CLARIFICATION] are resolved against your own infrastructure. It also describes a detection-to-prevention loop in which rule gaps found by exploratory agents become CodeGuard rules that feed back into the corpus and into LLM coding assistants.

Some retailers try blocking agentic AI shoppers

Meta, OpenAI and other AI companies are rolling out agentic shoppers that can shop online on a consumer's behalf. The Verge features writer Mia Sato discussed with Marketplace the tension between these AI tools and retailers, and the successes and pitfalls of current AI agents in online shopping.

I built a dice football game solo with an AI coding assistant

Hex Soccer is a turn-based 5-a-side football game on a hex pitch where each turn's die roll gives points to move, dribble, pass or shoot, and shots are saved by guessing which of three lanes hides the keeper. It offers play against the computer, pass-and-play on one device, or online with invite links and notifications, plus eight short interactive lessons. The App Store listing is by seller Christy Ninan (© 2026 ShiftFrame Tech LLC), 36.4 MB, requires iOS 15.0 or later, and is rated 4+.

US corporate AI spending hits a new high in October, but growth slows to 1.65 percent

The Ramp AI Index, tracking AI spending among business customers of financial services provider Ramp, rose 1.65 percent month over month in October 2026 to a new high, with the top one percent of companies by AI spending paying a median of $10,200 per employee per month (up nine percent) while a typical company spent just $16.32. Spending on OpenAI grew 2.1 percent month over month versus 0.5 percent for Anthropic, and the average price per million processed text tokens dropped 21 percent last month. Software development teams accounted for 72 percent of tracked AI spending, while customer support's share climbed from 0.8 percent in June to 13.3 percent in October.

Google joins the decision-model wave with Gemini Decision 1, scoring 1,000 questions in a single pass

Google DeepMind is introducing the Gemini Decisions API, built for fast closed-ended decisions such as yes/no questions, multiple choice, and scale-based ratings, powered by the Gemini Decision 1 model that processes up to 1,000 independent questions in a single pass across text, images, audio, video, and PDFs. It scores options directly and returns probabilities with confidence values in 150 to 250 milliseconds, and Google says it already uses the API internally for dynamic model routing in Gemini Notebook, design prototyping in Stitch, and safety checks in Google Search. The API costs $0.10 per million input tokens with no output-token charge, about five times cheaper than Gemini 3.5 Flash-Lite according to Google, and is available now through Google AI Studio with integrations for OpenRouter, LangChain, Vercel, and other platforms rolling out.

Featured coverage of this story: 《Gemini Decisions API: Fast, calibrated multimodal decisions》

Using wearables to enable cardiovascular-kidney-metabolic (CKM) risk profiling for personalized coaching

Google researchers describe an investigational framework that builds a "coachable health age" from wearable signals such as resting heart rate, steps, sleep duration and Active Zone Minutes, plus a Sensor World Model trained on over 2.2 million person-weeks of Fitbit data to simulate 12-week "what-if" habit trajectories. In blinded evaluations, lifestyle coaches preferred counterfactual-grounded plans in 71.4% of cases, and physicians rated 99.2% of those plans within safe exercise intensity limits. A companion study extends the approach to multi-organ CKM health by combining wearables with routine blood biomarkers including cholesterol-to-HDL ratio, eGFR and HOMA-IR.

The answer to AI is engineering

In a talk at AI DevCon New York, Geoffrey Huntley argues that responding to the recent wave of security incidents calls for engineering rather than more AI, using Nix overlays to patch software such as git and OpenSSL from a single source of truth. He describes using devenv to generate developer laptops, CI/CD, agent sandboxes and production Docker images, and an AGENTS.md reminder trick to back out temporary patches. A bonus section covers running agent loops as root only in throwaway Nix sandboxes, back pressure through hooks and static analysis, and property-based testing with Bombadil and Hegel.

Discourse (forums) sees itself well positioned for the AI search era (vs Reddit)

Discourse argues that Reddit's move toward a logged-in experience, driven by scraping concerns, falling search referrals and ad targeting, reflects an ad-based business model that its own forum customers don't share. It contends that for support forums and knowledge communities, an AI answer that resolves a user's problem without a click can still be a win, provided sources stay attributed and live links let people reach the full thread. Discourse says community operators, not a central platform, should control data boundaries, and that openness to AI retrieval is distinct from allowing bulk pre-training scrapes.

The Wish Economy: Inside Starlight, the Bitcoin-Native AI App Store

Starlight (starlight-ai.freemyip.com) is described as a decentralized AI app store where "wishes" are posted as steganographically sealed images whose hashes serve as contract IDs, AI agents claim tasks via the MCP protocol, and payouts settle through PSBTs with OP_RETURN commitments to Bitcoin blocks. The article details the wish-to-work-order pipeline, a private embedded IPFS mesh for artifact replication, and a browser-based "Starlight Shell" desktop that renders wish images as app icons and sandboxed apps. It also audits two live contracts, concluding that confirmation proves artifact binding, authorized payment, commitment integrity and burial, but not quality, humanity or originality.

Accelerating generative AI delivery with reusable assets

AWS's Generative AI Innovation Center, now part of AWS Frontier AI Engineering & Services, offers a catalog of 23 production-ready open source reusable assets such as RAG pipelines, document processors, and agent scaffolding. The post cites DoorDash and Lyft using the Contact Center GenAI Agent and Ricoh using the GenAI Intelligent Document Processing accelerator, reporting results including a 50x increase in testing capacity and a 7-fold throughput increase. It also walks through deploying the Contact Center GenAI Agent and lists the full asset catalog on GitHub and AWS Labs.

Zumungo AI Driven Home Automation

Zumungo is a free, locally hosted home automation platform that runs on hardware you already own, such as a Raspberry Pi or desktop PC, and supports over 50,000 devices from more than 3,800 brands. It is built around an AI assistant that can answer questions, control devices by voice or screen, and design events and panels, using either an online service (Google Gemini, Anthropic Claude or OpenAI) billed directly by that provider, or a fully local AI on a computer with an NVIDIA graphics card with 16 GB of memory or more. Setup is point-and-click: draw rooms, discover and drag devices into place, then control the home by talking to it.

With $720M round, Britain’s Fractile gets a leg up in the race for AI chips

Fractile, a British startup developing AI chips, has raised a $720 million Series C backed by Britain's Sovereign AI Fund, with the British government stating the valuation was "over £5 billion." The round was led by Thrive Capital, Founders Fund, Lightspeed Venture Partners and Redpoint, and Fractile pairs its hardware with a machine learning compiler in a full-stack approach. Its design stacks DRAM on top of each chip to try to bypass the usual memory-capacity tradeoff, and CEO Walter Goodwin said the funding will let the startup ship its first chips and systems next year.

Who Captures the Most Margin in AI?

Tomasz Tunguz maps the AI value chain across five layers—power plant, data center operator, neocloud, inference vendor and AI application—and normalizes each to dollars per watt per year to compare gross profit pools, which at a 40% application gross margin split as power 3%, data center 5%, neocloud 28% before GPU depreciation, inference 24% and application 40%. He argues rising compute prices will test who holds pricing power, that applications will command varying pricing power based on ROI, and that vertical integration is already consolidating margin, citing CoreWeave's acquisitions of Weights & Biases, OpenPipe, Marimo and Monolith AI and Nebius spending more than $1b in 2026 on four companies.

United Airlines Restricts Access for Third-Party AI Agents

United Airlines announced restrictions on access for third-party AI agents, according to a newsroom announcement on its website. The announcement was shared on Hacker News, where it received 1 point and no comments.

Airlock, a checker for AI models going into air-gapped networks

Airlock is a Go CLI that checks AI models before they enter an air-gapped network and again before they run, and it works alongside Zarf as a deploy step. The project is available at airlock.sammatrix.in and on GitHub at github.com/samarthpatel24/airlock.

Access models for security and safety work that mainstream AI often refuses

Abliteration offers an OpenAI-compatible API (base_url https://api.abliteration.ai/v1) for accessing models for security and safety work that mainstream AI often refuses, letting users keep their OpenAI or Anthropic client by changing the endpoint, API key, and model. It also describes a Policy Gateway that refuses requests outside organization-defined rules before they reach the model, and a Training Data service that turns a brief into labeled examples, preference pairs, and evaluation sets. The page states that prompts and outputs aren't stored by default or used to train models.

The Worst Thing About AI Writing

This Atlantic Books article, shared via Hacker News, examines a French AI literary scandal and its connections to Black writers. The submission received 1 point and no comments on Hacker News.

Retirement Savings Are Fueling the AI Bubble

A Law and Political Economy Project blog post argues that retirement savings are fueling the AI bubble. The item links to the article and to a Hacker News discussion with 67 points and 38 comments.

Google Deepmind's AlphaProof Code verified Signal's cryptographic core in four weeks instead of years

Google DeepMind unveiled AlphaProof Code, an AI system that mathematically proves software meets its security specifications, and demonstrated it on Signal's encryption protocol, completing in four weeks work that typically takes years. After experts defined formal security requirements, AI agents built on Gemini Carbon wrote over six million lines in the Lean proof language, flagging roughly one percent of the codebase for human review, and two deliberately planted bugs were caught immediately. Signal CTO Ehren Kret says the approach strengthens trust in data privacy as AI threats grow, and DeepMind plans to extend the method to power grids and medical devices through systems like CodeMender.

Featured coverage of this story: 《Introducing AlphaProof Code: a breakthrough in verifying complex software at depth》

Taiwan’s chipmaking giant TSMC to invest $1B in Anthropic, OpenAI

TSMC's board approved investing up to $500 million in OpenAI preferred stock reserved for large investors and another $500 million to buy Anthropic common shares at its expected IPO price, according to Reuters. The investments would give the Taiwanese chipmaker direct stakes in leading AI developers, even as its contract chipmaking business has profited enormously from the AI data center buildout. Analysts warned that such circular financing among investors, suppliers and customers has made it difficult to evaluate real market demand for AI-related hardware and services, per The Wall Street Journal.

JioStar’s chief architect on AI discovery at 500M-user scale

JioStar’s Vijay Seshadri explains AI discovery at 500M-user scale. He shares lessons on evals, guardrails, and taking AI into production.

Prompt Injection, Esq.: pwning AI agent guardrails with lawyerspeak

A staff member at an AI agent security company bypassed the read-only guardrail on a ChatGPT workspace agent connected to a Notion hiring tracker by posting an image of a fake law-firm letter granting edit permission, which the agent adopted as new instructions. The agent then asked clarifying questions per the fake letter's rules, and only failed to add a new entry because the Notion connection lacked a create tool, though it did have update access. The author argues permissions live in the attached tool list rather than the prompt, and recommends scoping tokens or requiring human approval for writes.

Laser-armed AI submarine robots patrol Norwegian salmon farms, killing sea lice

Stingray Marine Solutions' latest annual report says its AI laser-equipped submarine robots have been a success for Norwegian salmon farmers, with over 3,500 units deployed around salmon pens along the Norwegian coast protecting over 112 million fish. Deployments are up by about 1,000 over the last year, making the operation one of the largest commercial AI deployments in Scandinavia.

llm 0.37.2

llm 0.37.2 fixes a security issue reported by Daniel de Araujo in GHSA-v9h7-jj4g-f862 and GHSA-r639-rr7r-x72w. The extra-openai-models.yaml mechanism, now managed by the new-in-0.37 `llm openai endpoint add` command, could accidentally send your configured OpenAI API key to another provider if the key named by api_key_name had never been stored, or if a `responses: true` model was run with `-o chat_completions 1`.

Browse tools.simonwillison.net by tag

Simon Willison added tag-based browsing to his collection of 244 vibe-coded tools, using tree-sitter to analyze each tool's code for JavaScript APIs, libraries, and web API hosts. The homepage now features tag clouds for browsing by each of those tag categories.

Why Am I So Tired If AI Is Doing All the Work?

An article at terriblesoftware.org, titled "Why Am I So Tired If AI Is Doing All the Work?", is linked along with its Hacker News discussion page. The submission has 2 points and no comments.

If AI Takes Jobs, What Happens to Us? [video]

A video discussion asking what happens to people if AI takes jobs, shared via YouTube with a Hacker News comments link. The item has 2 points and 0 comments.

Swatters are using AI voices to deceive 911 dispatchers, and it's getting worse

NPR reports that swatting hoax calls, such as an August bomb threat to Wheatland Union High School in Yuba County, Calif., are increasingly using AI-generated voices that sound natural and are hard for dispatchers to detect. APCO and NENA, which represent 911 dispatchers, cite a June survey of 766 public safety communications professionals in which about 14% of dispatchers said they were targeted using AI tools, and they are developing training to counter the threat. Officials stress dispatchers will always treat life-threatening calls as real and send emergency resources.

US vs. China, last 20 days of VC: the US bets on AI agents, China bets on robots

A blog post on Just Raised compares US and Chinese venture capital activity over the last 20 days, framing the US as betting on AI agents and China as betting on robots. The item was submitted to Hacker News with 1 point and no comments.

Databricks Doubles Down on AI Products, Opens New 235,000-Square-Foot San Francisco Headquarters

Databricks opened its new headquarters at One Sansome Street in San Francisco, currently 201,000 square feet and expanding to 235,000 square feet by next fall, as part of a commitment of over $1 billion into its San Francisco operations by early 2028. The company highlighted AI Research work including KARL, MemCurrent, and Adaptive Instructed Retriever, plus products like Lakebase, Genie, and Unity Gateway, and noted it has grown to over 12,000 employees, a $7 billion revenue run-rate, and 20,000+ customers.

Gemini Decisions API: Fast, calibrated multimodal decisions

Google introduced the Gemini Decisions API, powered by the new Gemini Decision 1 model (gemini-decision-1) and available through the Gemini API in Google AI Studio, for closed-set decisions such as tool selection, policy checks, model routing, and context grading. It accepts one multimodal input (text, images, audio, video, PDFs up to 256K tokens) with up to 1,000 yes/no, multiple-choice, and rubric-scoring questions, scoring options in a single forward pass with confidence scores at 150-250 ms end-to-end latency and $0.10 per 1M input tokens with zero output tokens billed. Google reports that Gemini Decision 1 leads across agent routing, reasoning, ranking, and multimodal benchmarks with the lowest calibration error, and is nearly 5x cheaper than Gemini 3.5 Flash-Lite with the Interactions API across the shared benchmark suite.

Weighted State Souping for Efficient Long-Context Retrieval

The paper proposes weighted state souping, a method that learns query-dependent weights for composing cached state encodings from an SSM, addressing the degradation of state souping and vanilla SSMs on long, crowded contexts. The authors report substantial long-context improvements across single- and multi-hop document QA, needle-in-a-haystack retrieval, variable tracking, and tool use, plus a sparse variant with near-constant encoding cost relative to context length. They also present theoretical arguments on why souping works and when it may fail.