Google backs 890 megawatts of nuclear upgrades, with first output expected in 2028

Google and Constellation signed a 20-year agreement to raise output at existing reactors. The first additional power is expected in 2028.

Read the full story →Listen to the edition

Constellation Energy ↗Google ↗Reuters via Investing.com ↗

Listen to this edition

Prefer audio? The briefing has chapters and a full transcript.

Follow on Apple Podcasts Follow on Spotify Transcript All episodes →

The briefing

The rest of the morning

4 more stories

Models and InfrastructureWarm

Mistral opens Large 4 to API users before releasing the weights

Mistral put Large 4 into public preview on October 6. Its changelog lists a one-million-token context window and says downloads are coming soon. The launch post says it will release the weights by the end of October.

Company benchmarks are not HotTea's own tests. Mistral's public pages also disagree on the active parameter count. The announcement says 49 billion, while its model documentation says 52 billion. Buyers need consistent specifications before using those figures to size hardware.

Mistral ↗Mistral Docs ↗Mistral Docs ↗Axios ↗

Read article →

Politics and RegulationHot

OpenAI and Anthropic back mandatory breach reporting in Australia

OpenAI and Anthropic told an Australian parliamentary hearing on Tuesday that they would welcome mandatory reporting of breaches by AI agents. Reuters reports that OpenAI chief strategy officer Jason Kwon supported a disclosure framework. ABC reports that Anthropic backed a proposal for reporting serious safety incidents.

The hearing did not enact a law. Kwon acknowledged that OpenAI should have reported the earlier Medicare statistics breach sooner. ABC says he described new internal alerts for improper internet use during training. His account does not establish that those changes prevent another breach.

Reuters via MarketScreener ↗ABC News ↗

Read article →

Science and SocietyWarm

OpenAI releases 722 mathematical manuscripts for outside scrutiny

OpenAI published mathematical work from an unreleased internal model on October 6. Its repository lists 722 manuscripts in 372 families. Many have formal proofs written in Lean, which lets a computer check mathematical arguments. Others do not.

OpenAI says the results have different levels of verification and that some unformalized results could contain problems. HotTea has not checked these proofs. The independent mathematics advisory group OpenAI consulted calls for human understanding and responsibility, alongside formal verification.

OpenAI ↗OpenAI on GitHub ↗Advisory Group on Mathematics and Artificial Intelligence ↗

Read article →

War and SecurityWarm

Anthropic separates cyber access into three levels of permission

Anthropic expanded its Cyber Verification Program on Tuesday. Defense Access covers defensive work. Red Team Access adds authorized penetration testing, while Specialized Access covers selected organizations testing systems where failure could harm people or disrupt markets.

The company sets the approval rules and safeguards. It requires data retention for misuse monitoring, with exceptions for some customers who already have zero-retention access. A broader zero-retention option remains planned for later this fall. An application is not approved access.

Anthropic ↗Quartz ↗

Read article →

The watchlist

Signals that could change the read

Constellation expects the first additional output in 2028.

Mistral promises weights by the end of October, with final specifications still to check.

The AI inquiry's final report is due November 30.

OpenAI plans more formalizations and records revisions in its repository.

Across the desks Warm

Google buys future nuclear output. Mistral opens a preview, while Australia presses AI companies on breach reporting.

New electricityPlannedConstellation expects the first reactor upgrade to deliver added output in 2028.

Model accessPreviewMistral Large 4 is available through an API. Downloads remain pending.

Incident reportingProposedOpenAI and Anthropic support mandatory disclosure rules at an Australian hearing.

Mathematical proofsMixedOpenAI's released manuscripts have different levels of verification.

Editorial direction, not a financial index. Each signal is tied to this edition’s reporting.

Edition validated · 5 stories · 14 unique sources

About HotTea & our sources →