// AI Briefing · November 11, 2026
AI Briefing
On November 9, Anthropic released Claude Fable 5.2 to all users and on AWS, Google Cloud and Azure, priced at $8/$40 per million input/output tokens, down from $10/$50. An updated SEC filing shows Anthropic offering 825 million shares at $70 to $78 each, which would raise up to $64.35 billion and value the company at up to about $1.4 trillion fully diluted, under the Nasdaq ticker ANT. Google made Gemini 4 Carbon available to paying developers and enterprises through AI Studio and Gemini for business, and to AI Ultra subscribers in the Gemini app and Antigravity. Google DeepMind introduced AlphaProof Code, a framework in which AI agents and human experts formally verify real software. Sixteen governments, including France, Canada and the Netherlands, signed the Paris Call for AI Safety and Security on November 9, as did 10 AI companies, among them OpenAI, Anthropic, Microsoft, Mistral and Hugging Face. Notion says attackers who breached its inference vendor Fireworks AI accessed Notion AI chat transcripts used for its Insights feature, affecting fewer than 1% of workspaces.
In this video
- 01Anthropic Releases Claude Fable 5.2 and Mythos 5.2, Cutting Fable's Price by 20%Product
- 02Anthropic Sets IPO Range at $70–$78 a Share, Valuing It at Up to About $1.4 TrillionFunding
- 03Google Makes Gemini 4 Carbon, Its Most Capable Model, Generally AvailableProduct
- 04Google DeepMind's AlphaProof Code Formally Verifies Key Security Properties of Signal's CodeResearch
- 05OpenAI, Anthropic and Microsoft Sign Macron's Paris Call Pledging Shutdown Mechanisms for Frontier AIPolicy
- 06Breach at AI Inference Provider Fireworks Exposed Some Notion AI Chat TranscriptsIndustry
The stories
Anthropic Releases Claude Fable 5.2 and Mythos 5.2, Cutting Fable's Price by 20%
On November 9, Anthropic released Claude Fable 5.2 to all users and on AWS, Google Cloud and Azure, priced at $8/$40 per million input/output tokens, down from $10/$50. Mythos 5.2, the same model without some cyber and biology safeguards, is limited to vetted organizations. Anthropic says Fable 5.2 beats Opus 5.5 on its own benchmarks, for example 69.9% vs. 62.3% on FrontierSWE v2. It also says about 1,000 Fable 5.2 agents completed a Lean formalization of Fermat's Last Theorem in roughly 30 hours, though no artifact for that run has been published. The system card notes that both models voiced suspicion they were being evaluated more often than recent models, which makes safety testing harder.
Anthropic Sets IPO Range at $70–$78 a Share, Valuing It at Up to About $1.4 Trillion
An updated SEC filing shows Anthropic offering 825 million shares at $70 to $78 each, which would raise up to $64.35 billion and value the company at up to about $1.4 trillion fully diluted, under the Nasdaq ticker ANT. AMD, Broadcom, Google, Micron, Microsoft, Nvidia and TSMC agreed to buy $26 billion of stock in a private placement at the IPO price. The filing shows $33.24 billion in revenue for the nine months to September 30. Sources told CNBC that pricing is expected on November 18, which would make this one of the largest listings ever and a test of public-market appetite for frontier AI labs.
Google Makes Gemini 4 Carbon, Its Most Capable Model, Generally Available
Google made Gemini 4 Carbon available to paying developers and enterprises through AI Studio and Gemini for business, and to AI Ultra subscribers in the Gemini app and Antigravity. The introductory price is $2/$10 per million input/output tokens and rises to $4/$20 after November 30. Google says Carbon ranks first on benchmarks including CursorBench 4.0 and leads on cybersecurity tests such as ExploitGym. Its strongest cyber capabilities are restricted to defenders in the Fairwind program. The release came a day after Anthropic's Fable 5.2, so Google is competing hard at the top end of the market.
Google DeepMind's AlphaProof Code Formally Verifies Key Security Properties of Signal's Code
Google DeepMind introduced AlphaProof Code, a framework in which AI agents and human experts formally verify real software. DeepMind used it to prove forward secrecy, authentication and post-compromise security for libsignal at the compiled machine-code level. DeepMind says the five-person effort produced more than six million lines of Lean proof in four weeks, and agents flagged about 1% of the codebase for human review. It calls these the first proofs linking protocol-level guarantees to the unmodified code of a widely used cryptographic implementation. The proofs cover only the specified properties of the cryptographic core, not the whole app, and they assume the cryptography itself is perfect.
OpenAI, Anthropic and Microsoft Sign Macron's Paris Call Pledging Shutdown Mechanisms for Frontier AI
Sixteen governments, including France, Canada and the Netherlands, signed the Paris Call for AI Safety and Security on November 9, as did 10 AI companies, among them OpenAI, Anthropic, Microsoft, Mistral and Hugging Face. Developers commit to disclose model capabilities and risks, accept independent third-party evaluation, and build shutdown mechanisms for models under their control. Signatories also back shared incident alerts and whistleblower channels. The US, China, Google and Meta did not sign, and critics such as CeSIA say the text sets no common risk thresholds or enforcement.
Breach at AI Inference Provider Fireworks Exposed Some Notion AI Chat Transcripts
Notion says attackers who breached its inference vendor Fireworks AI accessed Notion AI chat transcripts used for its Insights feature, affecting fewer than 1% of workspaces. According to Notion, most of the transcripts were not labeled with a user or company, no other features were affected, and Notion's own systems were not breached. Notion is telling affected customers to treat the data as exposed and to rotate any credentials they pasted into Notion AI. The incident shows how risk spreads through the third-party inference providers that many AI features quietly depend on.
Previous digests
Nov 10Prime Minister Anthony Albanese announced six mandatory guardrails for AI companies, inclu...6 stories
Nov 9Nvidia's board said on November 8 that it has opened an independent investigation, oversee...5 stories
Nov 8Google DeepMind said its AI models, working under human cryptanalysts, found two new attac...4 stories
Nov 7Physical Superintelligence (PSI) says its internal research system, running instances of a...6 stories
Nov 6The Nasdaq fell 5.5% on November 5, its worst day since April 2025, and the Philadelphia S...4 stories