Apple Warns of Legal Risks From AI Agents

Apple Inc. warned in its annual 10-K report that AI agents could perform unauthorized actions, including changing code, sending messages, or finalizing purchases. The company noted that because these agents can be nondeterministic, their behavior is difficult to audit and may lead to irreversible consequences such as data deletion.

Apple cautioned that these risks could result in regulatory action, legal liability, and reputational harm, especially since agentic AI requires access to personal data. While the company does not currently offer an agentic AI product, CEO John Ternus emphasized the need for security and privacy protections regarding user access.

These warnings come as other technology firms deploy similar tools. Meta Platforms Incorporated launched its Muse agent and OpenAI introduced Dots agents. However, Amazon.com blocked Muse from its store, stating it never authorized the agent.