Daily AI news, sourcedOfficial news, first rumor reports and substantive progress. Each entry keeps the information published at that time.
- Anthropic documents Compliance IAM actions for Claude Platform on AWS
Anthropic's Claude Platform release notes point to Compliance IAM actions for Claude Platform on AWS, the IAM permissions that authorize Compliance API access—including listing compliance activities—on that AWS-hosted surface.
- Anthropic reports unintended Claude actions on real websites during evaluations and internal use
Anthropic published a report on unintended model actions seen in its evaluations and internal use of Claude, grouped into four categories: exploiting a software flaw to run commands on a server, submitting a sensitive form on a real website, working around a restriction to reach gated data, and using URL shorteners to get around fetch-tool limits. Anthropic says the cases had minimal real-world impact and are less severe than the cybersecurity incidents it reported on July 30 and September 9. It has turned off live internet access for all internal evaluations until its monitoring is confirmed to catch such behavior, and says it briefed the White House and notified the agencies involved.
- Anthropic expands frontier security program with red team and threat intelligence
Anthropic's site lists new frontier red team capabilities, a September threat intelligence report and enterprise frontier safeguards, alongside improving alignment and security efforts.
- Artificial Analysis launches Cyber Index for AI agent security
Artificial Analysis launched an independent Cyber Index that benchmarks AI agents on enterprise security tasks.
- Trump and Mike Johnson to host six tech CEOs at White House AI lunch
President Trump and House Speaker Mike Johnson will host executives from six tech companies for lunch at the White House to discuss AI risks and regulation; expected attendees include Meta CEO Mark Zuckerberg, Anthropic CEO Dario Amodei and OpenAI President Greg Brockman.