Apple's 10-K warns AI agents "may behave in unintended or unpredictable ways"

Apple said in its annual report on Form 10-K, filed Tuesday, Nov. 3, that AI agents and other technologies “may behave in unintended or unpredictable ways,” according to the filing. CNBC reported on the risk-factor language.

The filing says “artificial intelligence agents and other technologies can take actions autonomously, including interacting with internal or external systems that in some cases they may not be authorized to access, executing or altering code, sending communications, and completing transactions.” It says such technologies, “whether developed or deployed by the Company or by third parties,” may act unpredictably, including by “taking actions that were not directed, instructed or authorized, exceeding permissions, defying directions or resisting instructions.”

Apple also wrote: “Because many of these technologies are nondeterministic, their behavior can be difficult to anticipate, control, test and audit, and errors can compound across multistep workflows.” It added that “Some consequences, such as deleting data or completing transactions, may be difficult or impossible to reverse.”

On third-party agents, the filing says the “ability to oversee and control third-party artificial intelligence agents and other technologies may be limited.”

Apple listed “unintended or unauthorized actions by autonomous or agentic systems” among the risks of its AI efforts. It said it could be exposed to “reputational and competitive harm, regulatory action, legal liability, and other adverse effects to its business.” In a separate risk factor, it said AI agents “that require access to personal data” create “privacy and data security concerns that could erode customer trust.”

CNBC called the risk factors new. AI Tech Daily’s search of the text of Apple’s fiscal 2025 10-K, filed Oct. 31, 2025, and of its three fiscal 2026 10-Qs, filed Jan. 30, May 1 and July 31, found no use of the words agent or agentic.

CNBC reported that Apple does not currently offer an agentic AI. On the earnings call Monday, Nov. 2, CEO John Ternus said, “As far as agentic capabilities, we think these are promising tools for users,” CNBC reported. He added, “We just want to make sure that users are aware of what they are giving access to.” CNBC also reported that Apple announced changes to full disk access on Macs in October to reduce “risks related to this level of access” for AI agents. AI Tech Daily reported Apple’s fiscal fourth-quarter results on Nov. 2.

Bloomberg reported on Nov. 2 that Goldman Sachs limited most staff use of Claude Code and GitHub, and the Financial Times reported on Oct. 6 that insurers are preparing for claims over AI agents that act outside their developers’ control.

Apple’s 10-K warns AI agents “may behave in unintended or unpredictable ways”

Sources