Anthropic sets its IPO at up to $1.4tn, Claude gets Sheets
Anthropic's IPO range, Claude Sheets and Whiteboards, Gemini 4 Carbon, the Railway breach, DeepMind checking Signal, ChatGPT business profiles, Google's robots at Mercedes-Benz and New York's AI law.
Anthropic sets its IPO range at $70 to $78 a share At the top of the range Anthropic would be worth about $1.4tn, TechCrunch works out, and the share sale would raise up to about $64bn. Google and Nvidia are also buying $10bn of stock each at the IPO price, part of $26bn from chip and cloud companies. The new range came out on Tuesday, and trading should start around 19 November. Source
Claude now does spreadsheets and whiteboards, with meetings coming soon Sheets is a spreadsheet you and Claude build together, and Whiteboards is a shared board for laying out ideas with Claude. Both went into beta on all paid plans on Tuesday, and Claude Data, which used to be Dashboards, is out of beta. Claude Meetings, which will take notes on a call and update your documents as you talk, is coming soon. Source
Gemini 4 Carbon opens up, at half price until 30 November AI Ultra subscribers get it in the Gemini app, AI Mode in Search and Antigravity, and any paying developer can use it on the API at $2 per million tokens in and $10 out. Until Tuesday only a small group of cyber defenders had it, and the version tuned for cyber work still goes to approved defenders only. Source
Railway’s breach ran up about $740,000 in stolen AI usage Railway, a cloud platform developers use to host apps, says attackers broke in on 20 October and went mainly after the AI keys customers had stored there. Up to about 100,000 workspaces could have had stored secrets decrypted, according to the detailed incident report it put out on Tuesday. Railway says it’ll cover the token bills AI providers won’t refund. Source
DeepMind’s AI checked Signal’s encryption code in four weeks Agents running on Gemini 4 Carbon wrote more than six million lines of machine-checked proof that the core security of Signal’s actual code holds up, with only about 1 per cent of the code flagged for a human to look at. When DeepMind slipped two already-fixed bugs back into a copy, the system proved that version was insecure. It’s called AlphaProof Code, and DeepMind put the work out on Tuesday. Source
Businesses can now claim their own profiles inside ChatGPT Verified companies can update their product details and see how often ChatGPT mentions them, and some can set up a business agent that people chat to inside ChatGPT. It went into beta on Tuesday. OpenAI didn’t say how verification works, only that it differs by type of business and market. Source
Google’s robot model sorted Mercedes-Benz parts with 96 per cent success Gemini Robotics 2.5 ran Apptronik’s Apollo 2 humanoid, picking metal brackets out of bins and into trays, in a pilot on Mercedes-Benz production lines in Hungary and Germany. Google DeepMind put it out on Tuesday, and it’s rolling out to Google’s robotics partners. Source
Big AI labs must register with New York by 1 January The state opened the sign-up portal on Tuesday for AI developers making more than $500 million a year, ahead of its RAISE Act kicking in at the start of 2027. They’ll have to publish safety frameworks and report critical safety incidents within 72 hours. Fines run up to $1 million for a first violation and $3 million after that. Source
Newsletter
Get The Daily Debrief by email
The last 24 hours in AI, every weekday.
You're on the list.
One last step: tap the link we've just emailed you. It's how we make sure nobody can sign you up but you.