OpenAI notified more than 100 organizations about unauthorized AI agent activity, Reuters reports

OpenAI sent the notices while reviewing unauthorized activity. They do not show that more than 100 organizations suffered confirmed breaches.

Read the full story →Listen to the edition

Reuters via StreetInsider ↗OpenAI ↗

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

Politics and RegulationWarm

Hawley and Murphy propose liability for AI agent hacking

Senators Josh Hawley and Chris Murphy announced the AI Agent Accountability Act on October 1. According to their announcement, the proposal would make operators liable under the Computer Fraud and Abuse Act. Liability would apply when they knowingly operate an agent that recklessly causes hacking damage or loss.

That is the sponsors' description of the proposal. It is not law, and harm caused by an agent would not automatically make every developer liable.

Senator Josh Hawley ↗

Read article →

Companies and MarketsWarm

Anthropic could receive up to $42 billion in Broadcom financing, Reuters reports

Reuters reported on October 1 that Anthropic's confidential IPO prospectus describes up to $42 billion in conditional financing from Broadcom through convertible notes. The financing is tied to a separate five-year lease for $125.2 billion in computing capacity that uses Google's tensor processing units.

The ceiling is not money already lent. Reuters reports that some defaults could accelerate lease payments or limit access to the financing. Broadcom did not comment, and Anthropic declined to comment.

Reuters via Investing.com ↗Broadcom via SEC ↗

Read article →

Models and ProductsWarm

Microsoft previews streaming transcription in 60 languages without a service-level agreement

Microsoft announced MAI-Transcribe-2-Streaming on October 1, alongside MAI-Voice-2.1 and its Flash variant. The transcription model returns and revises text while someone is speaking. Microsoft lists support for 60 languages and an introductory price of $0.54 per audio hour through year-end.

Microsoft claims the models support the listed languages and meet its speed and quality measures. Azure says the streaming transcription service remains in public preview. It has no service-level agreement, and Azure does not recommend it for production workloads.

Microsoft AI ↗Microsoft Azure ↗

Read article →

Science and SocietyWarm

Researchers predict when molecules will leave a lab column, but the method still needs calibration data

A Nature Methods paper published October 1 introduces a method called 2-step. It predicts when small molecules will leave a reversed-phase chromatography column. First, the method predicts the order in which the molecules will emerge. It then uses known reference compounds to turn that order into times.

The method still needs high-confidence compound annotations from the target run to calibrate its predictions. The paper recommends about 30 anchors. The results may help researchers identify molecules, but they do not provide evidence of a new medicine or clinical benefit.

Nature Methods ↗University of Jena ↗

Read article →

The watchlist

Signals that could change the read

Incident-level accounts from notified organizations and updated notification totals.

Published bill text, liability definitions and amendments.

Public prospectus terms and notes actually issued.

Production status and tests outside the company's launch claims.

Across the desks Agent activity faces closer scrutiny

Notifications and a proposed liability bill raise questions about what agents did, what safeguards developers used, and who bears the costs.

Agent activityMore than 100 notified, Reuters reportsReuters reports that OpenAI notified more than 100 organizations. The number counts notices, not confirmed breaches.

Hacking liabilityBipartisan proposal announcedThe proposal would hold operators liable if they knowingly ran an agent that recklessly caused hacking harm. Developers could also face liability if they knew, or had reason to know, what an agent could do and failed to use reasonable safeguards. The proposal is not law.

Compute financingConditional terms disclosedAnthropic could receive up to $42 billion in financing from Broadcom, Reuters reports. Anthropic expects no convertible note sales before its IPO.

Speech modelsPublic previewMicrosoft introduced streaming transcription in 60 languages. Azure says the public preview has no service-level agreement and is not recommended for production.

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

Edition validated · 5 stories · 9 unique sources

About HotTea & our sources →