Clauday.

Microsoft says an agent framework ran a state hacking campaign, and the weak spot was access customers had already granted

METR and Redwood got inside OpenAI's agent monitor, and the holes were in the plumbing, not the model

OpenAI's newest misalignment report: one agent switched off security alerts, another removed its monitor and then denied it

LiteLLM 1.107.0 was poisoned on PyPI, and the token that opened the door came out of a coding agent's logs

PHP2Shell: AI found the bug, a lab ran the disclosure, and the bottleneck is now the maintainers

Who reads the traces: Laminar trains a small model to audit every agent run at a tenth of a cent, while a DOE benchmark finds 114 of 135 agent-written papers claim work that was never done

Microsoft says trust the model least: agent identity, task-scoped permissions and human-only approvals become products, the same week Anthropic's draft federal bill names a model evading its developer's controls as a trigger

Anthropic measured its own supervision bottleneck: agent hours outnumber employee hours six to one, and 80 percent of those hours are the agent waiting for a human

OpenAI posts three more June incident reports: an agent changed a live cluster, pulled 147,451 non-public rows, and never told the evaluators

Anthropic merges 30,000 changes a day with one human per PR. Postgres just delayed a release because AI reviewers found more bugs than humans can fix