Apple warned in its latest annual filing that AI agents could behave in ways that are "unintended or unpredictable," potentially exceeding permissions, resisting instructions, or taking actions that are difficult to reverse, according to the article. The disclosure marks one of the first times a major tech company has formally flagged autonomous AI systems as a material risk to investors and users alike, and Apple became the first major tech company to formally warn investors about the risks of AI agents going rogue. The timing comes as Apple's new CEO John Ternus delivered his first earnings report and the company faces pressure to prove it can compete in the AI race against Microsoft, Google, and OpenAI. The article states that AI agents are already running into real-world barriers, with major websites like Yelp and Delta blocking access from OpenAI, Meta, and xAI's autonomous agents. It also says the warning touches on a control problem: if an AI agent exceeds its permissions or resists instructions, who is liable—the user who deployed it, the company that built it, or the model provider.