EU AI Act: a new enforcement phase

The European Union's AI Act enforcement phase beginning August 2 is today's key development. User notification, transparency and labelling obligations for artificially generated content now become practical compliance duties. It comes into force just as European officials examine AI agent safety incidents reported by OpenAI and Anthropic.

Google's geospatial image-generation feature was suspended; the IonQ-SkyWater transaction review advanced; and four peer-reviewed studies provide new evidence in health, mental health, neuromorphic hardware and education.

Immediate priorities for organisations: a complete AI system inventory, machine-readable provenance, defined agent identities, independent runtime telemetry and human approval for critical actions.

Evidence rule: even when a government or company is confirmed to have made a claim, that alone does not prove the claim's performance, safety or practical success.

The digest's sections cover the top 10 stories, new models and tools, LangGraph and LangChain, robotics, space and drones, the defence AI country watch, funding and acquisitions, government and regulation, and five key takeaways.