Ten stories moved AI today, and money follows every one of them. Investors, regulators and a prime minister all placed new bets on where AI goes next. The biggest bet sits inside a legal filing. Anthropic just told federal regulators its own AI might be close to building a smarter version of itself.
Anthropic's Filing Raises the Stakes on Itself
Anthropic amended its IPO filing Monday with the Securities and Exchange Commission1. The filing warns Anthropic may soon cross its own threshold for recursive self-improvement. Crossing it means AI helps build a smarter version of itself. Anthropic defines that threshold as a doubling of its own pace of progress, caused by AI itself.
Daniel Kokotajlo took the warning seriously right away. He leads the AI Futures Project, the group behind the AI 2027 forecast. "It seems timelines to full RSI and superintelligence may be very short indeed," he wrote on X, hours after the filing became public2. His group reads a short path to superintelligence if the filing holds up.
Ross Hendricks reads the same filing with suspicion. He is a Porter and Company analyst who already ties Anthropic's biggest warnings to its biggest financial pressure. "Anyone still wondering why the company is threatening to end humanity unless the government shuts down their competition?" he asked in September3. An earlier leak of Anthropic's own numbers prompted the question.
Both readings share one fact. A federal filing carries legal liability; a blog post carries opinion. Anthropic itself admits it could miss the acceleration until after it happens, a limit the company wrote into its own warning1.
A Skeptic, a $1.4 Trillion Guess and a Filing Landed in 36 Days
Four dated posts and filings around Anthropic's amended IPO registration, Sept. 28 to Nov. 3, 2026
Each point marks one dated post or filing. The amended filing sits third, with a skeptic's post before it and a safety researcher's reaction the day after.
Sources: TechCrunch [1]; Daniel Kokotajlo on X [2]; Ross Hendricks on X [3]; Ed Zitron on X [14]. Chart by Ryan Elliott Dennis.
The numbers behind this chart
| Date | Event |
|---|---|
| Sept. 28, 2026 | Hendricks post. Ties Anthropic's safety warnings to its competition |
| Oct. 28, 2026 | $1.4T guess. Zitron pegs the IPO's low valuation estimate |
| Nov. 2, 2026 | Filing names RSI. Amended S-1A names the self-improvement threshold |
| Nov. 3, 2026 | Kokotajlo post. Calls the timeline to superintelligence short |
OpenAI Hires Uber's Voice to Calm Its Own
A job sat empty at OpenAI for ten months. Jill Hazelbaker will join as chief corporate affairs officer in December4. She leaves Uber, where she spent eleven years as president and chief corporate affairs officer.
Greg Brockman will be her boss. OpenAI's co-founder and president tied the hire to public trust itself. "As AI becomes part of more people's lives, how we communicate, engage, and earn public trust has never been more important," Brockman said4.
Hazelbaker built her career fixing trust problems. She ran communications at Snap and at Google before Uber hired her through its own string of scandals. OpenAI now bets her playbook works on a bigger stage.
Claude Code Loses Its Third Leader in Days
Cat Wu led product on Claude Code for two years. She posted on X Tuesday that Friday was her last day at Anthropic5. Her exit makes her the third person off that team to announce one inside two weeks.
Engineer Thariq Shihipar and desktop builder Anthony Morris left first, both on Oct. 245. Morris described his choice in his own words. "I created Claude Code on desktop because I believe everyone can be a builder," he wrote, calling the decision to leave "the hardest thing I've done in my career."
Anthropic kept silent on who leads the product next. Three departures from one product team, timed weeks before an IPO roadshow, read as more than coincidence.
cat@_catwuFriday was my last day at Anthropic.
Two years ago, when Claude Code was a small internal tool, I asked it to build an RL env. I steered it for tens of hours but never manually wrote a line of code. It worked, I was hooked, and I joined the team.
When we first launched, most…
A Shareholder Sues Alphabet Over Stolen Songs
Phil Rosen owns Alphabet stock, and he just sued its own board. His complaint names Sundar Pichai, Larry Page, Sergey Brin and ten other directors and executives6. He filed it Oct. 30 in federal court in Chicago.
The suit claims Google trained Lyria, Lyria 2, Lyria 3, Mulan and MusicLM on music it skipped licensing for. It points to 280,000 hours of songs pulled into MusicLM alone. Rosen argues the board chose speed over licenses on purpose, then stayed quiet about the risk in its own proxy filings6.
Rosen sues on Alphabet's own behalf, a derivative claim. Any money a court awards goes back to the company he says its own leaders put at risk.
Suno Asks a Judge to Drop Isbell's Claim
Jason Isbell and other musicians sued Suno over its AI vocals. They say the service copies a performer's identity, past any single song7.
Suno asked a federal judge in Massachusetts to drop the claims. Its lawyers argue copyright law protects a composition alone, apart from a performer's general style7. The company wants the case dismissed before a jury ever sees it.
The ruling will set a boundary line other voice-cloning suits will measure themselves against.
Microsoft Ties Three-Quarters of Its Budget to Chips
Last quarter alone, Microsoft spent $45.2 billion on capital equipment8. Finance chief Amy Hood explained the delay behind that number on the earnings call. "Capital expenditures were $45.2 billion, as some third party capacity deliveries we expected in Q1 were delayed," she said.
Roughly three-quarters of that total now sits in short-lived gear: Nvidia GPUs, CPUs and the memory that feeds them8. The short-lived share sat near half a year ago, then climbed to about two-thirds across the three quarters after it.
Chips that age inside a few years carry a different risk than a building that lasts decades. Microsoft is betting the payoff lands before the hardware wears out.
OpenAI Promises to Share the Power It Uses
A new OpenAI program goes by the name Grid Protectors. The pitch: let utilities cut an OpenAI data center's power first, before they touch anyone else's9.
Georgia gets the first deal. OpenAI will free up to 1 gigawatt at a campus it calls Project Camellia, under a 25-year deal with Georgia Power9. Kim Greene, the utility's chief executive, framed the trade in plain terms. "Data centers are paying more so families and small businesses can pay less, and our goal with every contract we sign under this framework is to deliver value for them," she said.
U.S. data center demand could pass 100 gigawatts by 2030. OpenAI wants its own campuses counted as part of the fix, apart from the strain.
OpenAI's Grid Pledge Covers 1 of the 100 Gigawatts It Expects
OpenAI's flexible-demand pledge in Georgia against the data-center demand it says the U.S. could reach by 2030
The ring is the 100 gigawatts of demand OpenAI expects by 2030. The magenta arc is the 1 gigawatt it pledged to free up at one Georgia campus, and the gray arc is the rest of that projected demand.
The 100-gigawatt figure is OpenAI's own nationwide projection for 2030, apart from any measure of its own fleet, so the ring sets one campus's pledge beside an industry-wide estimate.
Source: Data Center Dynamics [9]. Chart by Ryan Elliott Dennis.
The numbers behind this chart
| Item | Value |
|---|---|
| Pledged for curtailment in Georgia | 1 GW |
| Rest of 2030's projected demand | 99 GW |
| Whole | 100 GW |
Canada Buys Into Its Own AI Champion
Ottawa closed its stake in Cohere this week. The federal government put up to $1.6 billion into a $7 billion round that values the Toronto company at $35.7 billion11.
Prime Minister Mark Carney framed the money as a choice about power, past simple profit. "This is the moment that will decide where AI is built, how it is governed, and whom it serves. That's why we're investing in Cohere now. For affordable, secure, sovereign AI," he said10.
Cohere has to deliver on the bet. It must create at least 1,000 technical jobs and spend $1.7 billion on Canadian computing by 203011. The deal also keeps Cohere's headquarters inside the country.
Canva Cuts a Fifth of Its Staff
Layoffs at Canva will cut 1,271 jobs, about 22 percent of its staff12. Chief Executive Melanie Perkins called the past year the company's hardest yet.
Growth slowed for the first time in Canva's thirteen years. "After exploring a wide range of alternatives, I've come to the difficult conclusion that small changes won't be enough," Perkins wrote to staff12. She left AI unnamed as the direct cause.
Design tools built with AI now compete for the same customers Canva built its business on. The company that once automated design for everyone now trims its own team to answer that competition.
Congress Already Has a Bill for This
Rep. Ro Khanna wrote a bill before Anthropic's filing existed. His Human Control Over AI Act would ban models that improve themselves until a federal agency approves safeguards first13.
Khanna's bill would also build that agency from scratch, with power to license frontier models before launch and audit them after. He introduced it weeks before Anthropic named its own threshold in a legal filing.
The timing now reads differently. A bill written to stop a hypothetical case landed days before a company said the case might arrive on its own calendar.
By the numbers
- $1.4 trillion: the low end of the valuation range Anthropic's IPO is expected to price near14.
- Ten months: how long OpenAI's chief communications job sat open before Jill Hazelbaker took it4.
- Three: Claude Code team members who announced an exit inside two weeks5.
- 280,000 hours: music a lawsuit says trained Google's MusicLM alone6.
- 75 percent: the share of Microsoft's $45.2 billion quarterly capex tied to short-lived chips8.
- 1 gigawatt: power OpenAI pledged to free up at one Georgia campus during grid emergencies9.
- $1.6 billion: Canada's stake in Cohere's $7 billion round11.
- 1,271: jobs Canva is cutting, 22 percent of its staff12.
What to watch
Watch whether Anthropic names a model against its own threshold before the IPO roadshow ends this month. A confirmed crossing would move the self-improvement question from a filing's risk factor to a live event investors have to price. Watch Congress too. Khanna's bill sat with few co-sponsors in September, and a filing this specific could change that math fast.
Sources
- "Anthropic warns that it's very close to RSI capability, aka self-building AI," TechCrunch, Nov. 2, 2026, https://techcrunch.com/2026/11/02/anthropic-warns-that-its-very-close-to-rsi-capability-aka-self-building-ai/
- Daniel Kokotajlo, "Post on X," X, Nov. 3, 2026, https://x.com/DKokotajlo/status/2117738240757371022
- Ross Hendricks, "Post on X," X, Sept. 28, 2026, https://x.com/Ross__Hendricks/status/2104717983566176452
- "OpenAI hires Uber President Jill Hazelbaker to lead communications, filling a critical role that was curiously vacant for 10 months," Fortune, Nov. 3, 2026, https://fortune.com/2026/11/03/openai-hires-uber-jill-hazelbaker-to-to-lead-communications/
- Nitish Garg, "Cat Wu Leaves Anthropic: Third Claude Code Exit," CellCog, Nov. 4, 2026, https://cellcog.ai/blog/cat-wu-leaves-anthropic-claude-code/
- Mandy Dalugdug, "Alphabet shareholder sues Sundar Pichai and 12 others over training Google's AI, including Lyria, on copyrighted works," Music Business Worldwide, Nov. 2, 2026, https://www.musicbusinessworldwide.com/alphabet-shareholder-sues-sundar-pichai-and-12-others-over-training-googles-ai-including-lyria-on-copyrighted-works/
- "AI Co. Suno Seeks Toss Of Musicians' Claims It Copies Styles," Law360, Nov. 3, 2026, https://www.law360.com/pennsylvania/articles/2545975
- Gennaro Cuofano, "Microsoft Ties 75% of Its $45.2B Capex to Short-Lived Chips," FourWeekMBA, Oct. 28, 2026, https://fourweekmba.com/ai-microsoft-ties-75-of-its-45-2b-capex-to-short-lived-chips/
- Sebastian Moss, "OpenAI launches 'Grid Protectors Program' to reduce data center electricity demand when the US grid is under strain," Data Center Dynamics, Nov. 3, 2026, https://www.datacenterdynamics.com/en/news/openai-launches-grid-protectors-program-to-reduce-data-center-electricity-demand-when-the-us-grid-is-under-strain-openai-has-launched-a-plan-to-reduce-the-impact-of-its-data-centers-on-the-us-grid-the-non-binding-grid-protectors-program-aims-to-re/
- "Prime Minister Carney announces landmark investment in Cohere to scale secure, sovereign AI," Prime Minister of Canada, Oct. 29, 2026, https://www.pm.gc.ca/en/news/speeches/2026/10/29/prime-minister-carney-announces-landmark-investment-cohere-scale-secure
- "Ottawa spending up to $1.6 billion on AI startup Cohere," Yahoo Finance Canada, Nov. 3, 2026, https://ca.finance.yahoo.com/news/ottawa-spending-1-6-billion-120000780.html
- Samriddhi Srivastava, "Canva layoffs: 22% employees to exit, company offers six months' salary to affected staff," People Matters, Nov. 4, 2026, https://www.peoplematters.in/news/strategic-hr/canva-layoffs-22percent-employees-to-exit-company-offers-six-months-salary-to-affected-staff-53173
- Garrett Downs, "Khanna to introduce AI safety bill with ban on 'recursive' technology until safeguards exist," CNBC, Sept. 28, 2026, https://www.cnbc.com/2026/09/28/khanna-ai-safety-bill.html
- Ed Zitron, "Post on X," X, Oct. 28, 2026, https://x.com/edzitron/status/2115504670722474389
