Watchtower
Watchtower tracks when AI companies change their safety policies — and when they break them.
Adjusted authorship for a two-year-old article on their approach to alignment (with no substantive changes to the content)
OpenAI
Aug 18, 2024
Update 11/17/24: All credited authors on this article have now departed OpenAI.
On August 18, 2024, OpenAI revised the authorship of a two-year old article describing their approach to alignment research. The revision removed the name of Lilian Weng, who was the VP of research at the time, and replaced it with the recently-departed John Shulman (the former head of alignment).
No other substantive changes were made to the document, and OpenAI did not respond to requests for comment.
Meta
Oct 2, 2026
Meta updates and renames its safety framework from the "Advanced AI Scaling Framework" (v2) to the "Meta Superintelligence Scaling Framework" (v2.1).
Anthropic
Sep 22, 2026
Anthropic’s release of Opus 5.5 with an “inconclusive” determination for harmful-manipulation.
xAI
Sep 21, 2026
xAI says Grok 4.7 scores below its safety framework’s capability thresholds on dual-use knowledge, but xAI’s safety framework doesn’t provide thresholds.
OpenAI
Sep 9, 2026
Updated the GPT-6 Astra system card, specifically to hedge claims about misalignment.
OpenAI
Aug 18, 2026
OpenAI updated its Model Spec, the document outlining intended model behavior
xAI
Aug 17, 2026
xAI updated Grok 4.6's model card after release. A changelog was included.
Frequently asked questions
Have more questions? Our team is happy to help, contact us.