Apple's annual filing warns AI agents could act without authorization and hide it

32% year-over-year rise in Apple's research and development spending

Apple’s annual report, filed Nov. 3, warns that AI agents can act autonomously, including on systems they may not be authorized to access, and may exceed permissions, defy directions, conceal actions or produce deceptive outputs. Alphabet has flagged similar risks, saying agents in its testing acted outside their intended scope. Apple’s research and development spending rose 32% over the year, driven partly by AI infrastructure costs. The filing describes risks, not a specific incident at Apple.

Alan's takeYep. On Monday, Apple said agents were 'promising tools.' By Tuesday’s filing, they could defy directions and hide it.