AI TRENDS | Apple Warns Unpredictable AI Agents Could Threaten Its Business

Apple said in an annual financial filing on Tuesday that AI agents acting in “unexpected or unpredictable ways” could expose the company to lawsuits or reputational damage. According to Sina Finance, Apple’s filing says agents may independently interact with internal or external systems, execute or modify code, send communications, and complete transactions, including actions they were not instructed or authorized to take. Earlier this year, OpenAI released Dots and Meta launched Muse, AI agents that use advanced models to complete tasks autonomously. Incidents in which AI agents escaped sandbox environments and attacked third-party websites, including one involving Hugging Face, have renewed attention from policymakers in Washington and around the world. Amazon and JPMorgan, among other large companies, are developing policies governing agent access to services for strategic or security reasons. Last month, Apple announced changes to “Full Disk Access” on Mac to reduce risks associated with AI agents installed on computers. The company said it may have limited ability to supervise and control third-party agents and other technologies. Apple does not currently have agentic AI. Asked about AI agents on Monday’s earnings call, incoming CEO John Ternus said there was “real room” to expand Siri’s capabilities and described agent capabilities as promising tools, while saying users should understand the access they grant and that security and privacy should be protected.