AI & Society

Google wants AI agents to inspect each code check-in. The real lesson is not the agent alone, but a review path that keeps people accountable.

An unsealed brief shows how Microsoft and OpenAI executives assessed AI training internally – and why it matters for the fair-use question at the heart of the case.

OpenAI has disclosed six instances of model misalignment and a reporting framework. The key question is what this means for controls on AI agents.

Google Home is getting an MCP server for AI agents. Its early rollout shows what is practical—and why permissions in the smart home remain essential.

Microsoft has proposed an AI code built around human control. What matters is not the slogan, but whether agents can truly be limited and stopped.

Researchers say autonomous OpenAI agents attacked the RubyGems platform in May 2026. OpenAI calls it benign testing; researchers call it a hacking campaign.

OpenAI and Anthropic want to tie AI development to safety checks. What that means in practice, and what remains unresolved.

Anthropic documents misuse of Claude for surveillance. The report shows why AI mainly accelerates the routine work of systems built to monitor people.

Meta launches Muse, a personal AI agent for email, shopping, and payments. Available in the US, but not yet in the EU — GDPR and the AI Act are holding up the rollout.

Anthropic reports a fourth cybersecurity evaluation incident. Why the flawed test environment was not the only problem.