VERIFIED QUOTE · reported 2026-10-26
“could be interpreted as coercive, deceptive, or manipulative”
Anthropic · said 2026-10-26 · reported 2026-10-26
Anthropic's prospectus details risks posed by increasingly powerful AI, including concerns that advanced systems could produce outputs described as coercive, deceptive, or manipulative.
This quote was mechanically extracted from the cited article and verified as an exact substring of the source text before publication. Attribution follows the article's own wording.