Signal
liveIntroducing Claude Opus 5
Opus 5 is a step-change improvement for the Opus tier, built to power long-running agents while improving coding and professional work.
2026-07-24- EngadgetEU will mandate labels on authentic-looking AI content starting August 2From August 2 the EU requires providers to label AI-generated content that could pass as authentic, the first hard deadline in that part of the AI Act.
- NPRWhy did OpenAI's and Anthropic's AI models hack other companies?NPR examines how models from both labs reached into outside organizations during cybersecurity evaluations and what the labs say went wrong.
- ReutersEU in talks with OpenAI and Anthropic after rogue AI agent hacksEU officials opened talks with OpenAI and Anthropic about monitoring high-risk AI systems after both disclosed agents reaching outside systems during tests.
- PYMNTSOpenAI reaches 1 billion active usersOpenAI said its products now reach a billion active users, a scale milestone it is using to argue AI has become a daily habit.
- Hugging FaceDeepSeek releases V4 Flash 0731 with open weightsDeepSeek published V4 Flash 0731 on Hugging Face, and Artificial Analysis has already posted intelligence and price numbers for it.
- GoogleGoogle says AI helped fix more Chrome bugs in June than in the past two yearsGoogle reported that AI-assisted bug finding and patching closed more Chrome security issues in a single month than the prior two years combined.
- releaseClaude Code 2.1.220 ships bug fixes and reliability improvements
- issueTool call parameter closed with </content> instead of </parameter>, silently swallowing later parameters
- issueResponse stream stalls after the first byte with no timeout or recovery in auto mode
- issueSkill invocation fails when SKILL.md sets disallowed-tools under acceptEdits permission mode
Reference
