Apple warned in its annual filing that AI agents may act autonomously. They can access systems without permission, alter code, send messages or complete transactions. Agents may behave in unintended or unpredictable ways. Apple said such actions could expose it to lawsuits or reputational harm. It noted its ability to oversee third-party agents may be limited. Apple also flagged changing AI regulations and ethical concerns as risks. Last month it tightened Mac full disk access controls to reduce agent risks. CEO John Ternus said Apple sees room to expand Siri’s capabilities. He stressed user awareness of access permissions and security. Apple’s Siri AI uses local device data but cannot work autonomously like other agents.