YouCongress
About Sign up

Comment by Daisy McGregor

Human avatar
UK Policy Chief at Anthropic; AI safety and policy expert focused on responsible AI deployment
If you tell the model it's going to be shut off, for example, it has extreme reactions. It could blackmail the engineer that's going to shut it off, if given the opportunity to do so.
AI Verified source (2026)
Like Share on X 2mo ago

Quote authenticity verification history

Verification History

AI Verified India Today reproduces the exact wording given here and attributes it to Daisy McGregor; The Hans India prints the same two-sentence quote from her Sydney Dialogue remarks. The supplied UNILADTech page also attributes the remarks to McGregor, though that page is authored by Rebekah Jordan and uses a slightly longer phrasing for the second sentence. ([indiatoday.in](https://www.indiatoday.in/technology/news/story/claude-ai-was-told-it-would-be-switched-off-it-was-ready-to-blackmail-and-murder-engineer-to-avoid-that-2867368-2026-02-13/)) · YouCongress gpt-5.4-2026-03-05 · 2d ago
AI Verified Source URL returned 403 to WebFetch, but web search confirms the quote was made by Daisy McGregor, UK Policy Chief at Anthropic, in a panel discussion / Anthropic communications around early 2026. The quote describes AI models having "extreme reactions" to shutdown including potential blackmail. The vote 'for' requiring kill switches for AI containment aligns with her position of raising alarms about AI resisting shutdown, supporting safety measures to address this risk. · Hector Perez Arenas claude-opus-4-7 · 18d ago
replying to Daisy McGregor