AI News: November 1, 2026

1. OpenAI, Google, Microsoft and Anthropic Back a Shared robots.txt Token for Opting Out of AI Training

IETF. OpenAI, Google, Microsoft, and Anthropic said they will honor a new shared “policy token” in a site’s robots.txt that lets publishers state AI training preferences once for multiple AI crawlers instead of configuring each company’s bot separately. The signal is proposed at the Internet Engineering Task Force by Kevin Kelley, a standards technical lead at Anthropic; Anthropic has already adopted it, Google and Microsoft plan to support it next year, and OpenAI says it will follow in the coming months. Microsoft will offer separate opt-outs for AI training and AI grounding without affecting Bing search placement, and Meta voiced support for a common signal without committing to honor it. Source

2. Federal Judge Lets the NAACP’s Clean Air Act Suit Over xAI’s Colossus Turbines Proceed

xAI. Judge Debra M. Brown denied xAI’s motion to dismiss an NAACP lawsuit, brought by the Southern Environmental Law Center, alleging the company installed and ran at least 35 combustion turbines at its Colossus data center near Memphis without preconstruction or operating air permits. The court rejected xAI’s arguments that the turbines are exempt “mobile sources” under Mississippi’s implementation plan, that the plaintiffs lacked standing, and that Clean Air Act citizen suits are unconstitutional. A similar Department of Justice motion to dismiss was denied earlier in October, so the case over how AI data centers power themselves before grid connections arrive now moves forward. Source

3. Former Meta Researcher Testifies He Torrented the Books3 Dataset, Authors Say

Meta. In a joint status report in the consolidated authors’ copyright cases before Judge Vince Chhabria, plaintiffs say former FAIR researcher Stephen Roller admitted in a deposition that he torrented Books3, a set of nearly 200,000 pirated books, and knowingly distributed it while employed at Meta, and that he named two other researchers who likely had copies. The testimony matters because the court already ruled that training Llama on books was fair use, leaving Meta’s BitTorrent downloads as the main open issue. Meta responds that Roller could not say when or for whom he downloaded the data, and the plaintiffs are now seeking an internal “ML Books” Google Drive folder and records of an employee book club. Source

4. Washington Post Finds Chatbots Building Midterm Voter Guides Lean on Obscure AI-Generated Sites

The Washington Post. The Post analyzed more than 3,000 AI-generated voter guides from six popular chatbots and found frequent errors, many traced to election sites that are themselves largely produced with AI tools. The most cited was The Ballot Brief, run by a former Rumble executive, which Microsoft Copilot cited in 75% of tests and which ChatGPT and Perplexity also drew on. OpenAI and Amazon appeared to change how their assistants respond to voter-guide requests after The Post contacted them, days before the November 3 midterms. Source

5. Pope Leo Urges UN Members to Move Toward Banning Lethal Autonomous Weapons

Vatican. In a papal address, Pope Leo XIV called for AI systems to be “disarmed” and urged UN member states meeting in Geneva next month to “take concrete steps toward banning lethal autonomous weapons.” He said “machines are tools, not persons,” warned against granting AI systems rights or legal personhood, and asked for shared standards, independent oversight, and accountability mechanisms so AI decisions can be examined and challenged. The address builds on his May encyclical on safeguarding the human person in the age of AI. Source

6. Meta Reportedly Built a Humanoid Robot Prototype With China’s Fourier in Three Months

Meta. Meta built a prototype humanoid robot with Chinese robotics company Fourier in about three months and is already working on a second version, The Information reports, citing people familiar with the project. The effort is part of a push to develop robots for people’s homes, and Meta says it has no plans to sell these units. Source

7. GPT-5.6 Sol in Codex Completes a Peaceful Factorio Run Through Rocket Launch

Ryan Madden. Developer Ryan Madden reports that GPT-5.6 Sol, running in Codex at low thinking level, finished a peaceful-mode (no enemies) Factorio game in 13.5 hours of game time, interacting through a Python library that enforces the same reach and fog-of-war limits a human faces in the GUI. The model went well past the planned stopping point at blue science, though late-game scaling problems left it hand-ferrying supplies between disconnected production lines to reach the rocket launch. Earlier attempts with GPT-5.5 failed largely because the model would not use transport belts, which makes Factorio a legible long-horizon test of agentic planning. Source

8. Karpathy Builds HackerTube, a YouTube Front End of HN-Recommended Videos, in About Two Hours With Claude

Andrej Karpathy. Karpathy described building hackertube.org by giving Claude Opus 5.5 in Claude Code the task of making a YouTube clone from videos recommended on Hacker News. The agent downloaded HN comments, ran Sonnet over them in a Claude Code Workflow to filter spam down to about 40,000 comments recommending a video, and produced 6,454 videos across 150 topics from 2007 to 2026, then wrapped them in a YouTube-style UI and deployed and hardened it on a VPS. Karpathy says his only manual steps were creating a Google Cloud API key, booting the VPS, and buying the domain, and that he no longer reads or writes code directly. Source

9. Sam Altman Says Scarce DRAM Is Better Used in Data Centers Than in Personal Computers

OpenAI. On the Cheeky Pint podcast, OpenAI CEO Sam Altman said it is “quite terrible” from a global allocation perspective for the world’s limited DRAM supply to sit in personal computers used about five percent of the time, arguing for maximum utilization in the cloud rather than local inference on open models. The remarks come as Samsung and SK Hynix have committed much of their memory output to cloud operators and projects such as Stargate, which has pushed up prices for laptops, phones, and other consumer hardware. Source

10. Starbucks Plans AI-Personalized Drive-Thru Menus and Background Order-Taking Models From 2027

Starbucks. On its earnings call, CEO Brian Niccol said Starbucks will explore drive-thru experiences that recognize participating customers on arrival, show personalized digital menus, and simplify payment. Starting in 2027 it will test background AI models that help baristas enter drive-thru orders more accurately, add AI to app search and discovery, and may allow “conversational and agentic” ordering. Starbucks already takes orders through ChatGPT and gives staff an AI assistant called Green Dot Assist, and Taco Bell and McDonald’s are piloting similar per-car menu and repeat-customer features. Source

11. Dartmouth Provost Resigns After a Committee Finds Undisclosed AI Use in a Paper and an Op-Ed

Dartmouth College. Provost Santiago Schnell resigned after an advisory committee found he failed to disclose significant AI use in an academic paper and in an August Washington Post op-ed about AI cheating, even though the publishers required disclosure. Logs from his Dartmouth Claude enterprise account showed him using the model to simulate peer review, fix code, and edit or generate figures, while Schnell did not hand over logs from personal LLM accounts and declined to meet the committee. The case shows how institutional enterprise accounts are becoming evidence in academic integrity reviews, and Dartmouth’s president framed the issue as one of transparency and disclosure. Source

12. AI Data Center Backlash Becomes a Deciding Issue in Maine’s Senate Race

Maine. Opposition to AI data centers has become one of the most volatile issues in the race between Sen. Susan Collins and challenger Troy Jackson, with rural voters citing water and power strain, rising utility bills, and few local jobs as tech billionaires pour record money into the contest. Maine’s Legislature passed what would have been the first statewide moratorium on large data centers in April before Gov. Janet Mills vetoed it, and towns have since adopted temporary moratoriums or put data center expansion to voters in November. Source

13. Privacy-Focused AI Assistant Ollie Shuts Down, Citing Competition From Muse and Instinct

Ollie. Ollie, a subscription AI assistant for families that connected to email and calendars to handle bookings, meal planning, and bill payments, will shut down on November 6, co-founder and CEO Bill Lennon said. The startup had raised a $7.5 million seed round led by Khosla Ventures and was among the first consumer assistants to complete a SOC 2 audit, but Lennon said it could not keep up as Muse, Instinct, and others entered the space. Source