Apple's annual report lists unpredictable AI agents as a business risk

Kyunghyang Shinmun reports that Apple listed the possibility of unexpected actions by AI agents as a business risk in its fiscal 2026 annual report, filed with the US Securities and Exchange Commission on November 3. In the filing, Apple says that "AI agents and other technologies can act autonomously," including by interacting with internal and external systems, running and modifying code, sending communications and completing transactions. Apple warns that such agents, whether built and deployed by Apple or by third parties, may act without human instructions, commands or approvals. They may not follow instructions, may behave unpredictably, may get around security and control measures, and may be exploited by malicious attackers. In those cases, Apple says, errors could build up in processes that use AI, data could be deleted and other irreversible outcomes could follow. The filing also points to uncertainty over laws and regulations covering AI and superintelligence, and says negative consumer perceptions and ethical concerns could weigh on its business. Risk-factor sections in annual reports are meant to disclose possible harms, not predict them. Still, the language shows that a major consumer platform company now treats autonomous agent behavior as a material risk. The filing came as Apple is opening its Siri AI assistant to third-party apps.