Skip to content

Technology coverage

AI safety and autonomous systems

Where systems behave in ways their operators did not intend, and what the evidence actually establishes.

Newsroom focus

Peer-reviewed and preprint safety research, evaluations, documented agent and autonomous-system behavior, and risk analysis attributable to named researchers or institutions.

Publisher coverage / live source wire

AI safety wire

Go to our reporting ↓

Headlines matching AI terms together with safety, risk, evaluation or autonomous-system terms. These matches are not verified incidents. Headlines arrive from direct publisher feeds and Google News; open the source to read. No generated articles or summaries.

Saved stories (0) ↗

Checking publisher feeds…

Feed checks and how headlines are selected

8 geographic Google News feeds and 5 direct publisher feeds load independently; local publisher results can arrive first. Results use headline words for geography and topics, and may miss or misclassify coverage. The Google search window is 14 days; direct feeds contain the publisher’s selected recent items; older returned records keep their publisher dates. Matching titles from the same publisher appear once, assigned to the most local matching feed. Refreshing never changes a publication date. Known press-release services are labeled; other outlets can also carry sponsored material. External coverage has not been independently verified by our newsroom.

Pass AI Watch / editor-reviewed work

Our reporting

Original reporting published by our newsroom, with named authors and sources.

Launch state

No safety coverage has been published yet.

This collection carries a hard rule: we do not describe a hypothetical failure in language that suggests it happened. Research findings, evaluations, and documented incidents are labeled distinctly.