← Daily AI news

Daily AI news, Wednesday, 4 November 2026

Daily AI / Tech / VC Briefing: Wednesday 4 November 2026

The US Capitol building in Washington, DC
Photo: Wikimedia Commons

Good afternoon, here's today's briefing. Voter anger at AI data centres helped Democrats take both houses of Congress, OpenAI is buying a developer-tools startup to take on GitHub, and one of the leaders behind Claude Code has left Anthropic weeks before its IPO.

LEAD: AI's first big election goes badly for the industry

Background: Yesterday's briefing previewed the midterms as the first real test of AI as a political issue, with rival industry-funded super PACs on both sides.

What happened: Democrats won control of both the Senate and the House. TIME reports that an NBC exit poll found 56% of voters called AI and data centres a "crucial or major factor" in their vote, and Democrats spent more than $100 million on ads tying Republicans to Big Tech and data centres. Sherrod Brown won Ohio's Senate race after branding his opponent "the face of data centers in Ohio," and data-centre bans or moratoriums passed in counties across the country. In Texas, which Wells Fargo says holds 27% of planned US data-centre capacity, Democrat Gina Hinojosa won the governor's race on a promise of an immediate data-centre moratorium. Elon Musk, the cycle's biggest donor, saw his backed candidates in Texas lose. SpaceX shares fell about 4% after the open and Tesla about 5%, along with AI infrastructure stocks such as CoreWeave; Musk posted that "data centers in space" were "looking pretty appealing right now." The pro-AI super PAC Leading the Future made little dent, while candidates backed by pro-safety groups won several races.

Why it matters: Sweeping federal AI law is still unlikely with a presidential veto in play, but expect tougher hearings, more state-level rules, and slower permits for data centres, especially in Texas. OpenAI is already reported to be rebranding data centres as "compute campuses."

Sources: TIME · CNBC · CBS News · Politico

OpenAI buys Pierre to take on GitHub

Background: Coding agents now write so much code that the tools for storing and reviewing it are under strain. OpenAI recently described Pando, its in-house source-control system, which it says cut the time to push code changes by nearly 80%.

What happened: OpenAI agreed to buy Pierre Computer Company, maker of the Code Storage service and the open-source Trees and Diffs libraries already used for browsing files and reviewing changes in the Codex app. Terms weren't disclosed. Pierre's open-source projects stay open, and the deal still needs regulatory approval. Axios framed the deal as a direct challenge to Microsoft's GitHub. Separately, OpenAI said searches inside ChatGPT now equal about 20% of Google's search volume, and that weekly users grew from 1 billion to 1.2 billion in 60 days, though it didn't say how it measured Google's side.

Why it matters: OpenAI wants Codex to own the whole path from an agent's first branch to deployed code, not just the writing. That puts it on a collision course with GitHub and with every coding tool, Cursor included, that relies on GitHub today.

Sources: OpenAI · Axios · The Verge · Big Technology

OpenAI staff warned for years about weak security, Reuters reports

Background: A former OpenAI engineer, Abdulrahman Alfozan, has been indicted for allegedly hiding an audio recorder in an executive conference room.

What happened: Reuters, citing five people, reported that employees warned leaders and the board for years, some before ChatGPT launched, that weak controls could expose sensitive technology, and that some came away believing management cared more about keeping pace with Google and Anthropic. One concern was broad staff access to a shared code repository. OpenAI has raised its top bug bounty to about $172,000 and signed a security partnership with Cathedral Systems, a startup founded by former DOGE staff that previously got into parts of Anthropic's systems during a bug-bounty test.

Why it matters: Ahead of a possible IPO and with a new Congress keen on oversight, security at frontier labs is becoming a governance issue, not just an engineering one.

Sources: AI Affairs, summarising Reuters · Chosun Biz

Anthropic: a Claude Code leader leaves as the IPO clock runs

Background: Anthropic is preparing to list on Nasdaq under the ticker "ANT," in what is expected to be one of the biggest IPOs ever.

What happened: Cat Wu, head of product for Claude Code and Cowork and, alongside creator Boris Cherny, one of the tool's two best-known leaders, left Anthropic on Friday and announced it on Tuesday. She gave no reason, and Anthropic confirmed the exit. Business Insider notes that Thariq Shihipar, who also worked on Claude Code, left last month. On timing, Yahoo Finance calculates that SEC rules allow the roadshow to start from 10 November and trading from 16 November at the earliest; nothing is confirmed.

Why it matters: Claude Code is the product that turned Anthropic into a real rival to OpenAI, so leadership churn there, weeks before a giant IPO, will be watched closely by investors.

Sources: Business Insider · Analytics India Magazine · Yahoo Finance

Competitor moves: big customers push back on coding-agent costs and risk

Background: Coding agents are now one of the largest AI line items for big companies, and buyers are starting to shop around or tighten access.

What happened: Revolut says it cut its AI coding bill by about 90% in under a year by making Z.ai's open-weight GLM model, self-hosted on 56 Nvidia B200 chips, the default for roughly 2,000 engineers, at about $0.30 per million tokens. Claude Code and Codex are now kept for exceptions, and image work is routed to other models. Its infrastructure lead said cheap, flat-rate tokens let engineers hand agents debugging and ticket work they used to ration. Separately, Bloomberg reported that Goldman Sachs has restricted Claude Code and GitHub tools to engineers, cutting off most traders and bankers apart from its quantitative "strats." Elsewhere, GitHub Copilot arrived in JetBrains' new Air agent environment.

Why it matters: This is real price pressure on Anthropic, OpenAI and every coding-agent maker, including Cursor: once open models are good enough, large customers will run them in-house and pay frontier prices only for the hardest jobs.

Sources: The Register · Bloomberg · The New Stack

Chips and big-tech money: AMD surges, Meta borrows

What happened: AMD's third-quarter revenue rose 45% to $13.38 billion, with data-centre sales up 97% to a record $8.55 billion, now 64% of the business; shares jumped about 8% after hours. Meta sought $20 billion to $25 billion in bonds to fund AI and priced $15 billion of dollar notes, with its 30-year bonds paying 7.5% on up to $58 billion of orders.

Why it matters: Demand for AI chips is still strong, but even the richest companies are now borrowing heavily, and at higher rates, to keep building.

Sources: AMD · CNBC on AMD · CNBC on Meta · 24/7 Wall St.

VC and deals

What happened: Fal, which runs image, video and audio models for developers, raised a $1.3 billion Series F led by Bessemer at a $16 billion valuation, double its April price; Bloomberg's source puts annualised revenue at $800 million. Traversal, which builds AI agents that find and fix production outages, raised a $120 million Series B at a $1 billion valuation, co-led by Kleiner Perkins, Redpoint and Khosla. AI sales start-up Nooks secured $500 million from General Catalyst's Customer Value Fund, which takes a share of returns rather than equity, and launched Node, a sales assistant. Sentra raised $150 million, led by Capital One Ventures, to control which company data AI agents can reach.

Why it matters: Money is flowing to the picks and shovels around agents: running models, fixing what agent-written code breaks, and fencing in what agents can touch.

Sources: TechCrunch · Tech.eu · Business Wire · Reuters via Investing.com · SiliconANGLE

Hires and departures

The biggest people move at a coding-agent company today is Cat Wu's exit from Anthropic's Claude Code team (above). No other notable hires at Cursor, Cognition, OpenAI's Codex team or other coding-agent companies surfaced in the last day.

← Tuesday, 3 November 2026 Thursday, 5 November 2026 →

Get the weekly sift

The best new AI and agent releases, with one-line verdicts. Occasional picks, no spam.

Privacy policy: your email goes to our email provider, Resend. Unsubscribe any time.

News is editorial and never paid for. Sponsored content on ToolSift is always labelled.