Apple has expanded its warnings about the risks of artificial intelligence, cautioning that autonomous AI agents could access systems without authorisation, send communications, execute transactions and behave in unpredictable ways that may threaten businesses and users, as reported by Fortune. The warning was included in the iPhone maker’s latest annual 10-K filing, released a day after the company reported its fourth-quarter financial results. It said AI agents and similar technologies can act on their own: interacting with internal or external systems they may not be authorised to access, executing or altering code, sending communications and completing transactions. Such tools, Apple said, may behave in unintended or unpredictable ways. That includes taking actions nobody directed or authorised, exceeding permissions, or defying and resisting instructions. The company also said rogue agents could conceal their actions, withhold or manipulate information, or generate inaccurate or deceptive outputs.
Apple highlights risks from autonomous AI agents
In the risk factors section of its annual filing, Apple said AI agents and similar technologies are capable of taking actions independently, including interacting with internal and external systems, executing or altering code, sending communications and completing transactions. The company warned that such systems may sometimes attempt to access platforms or services they are not authorised to use. Apple said these technologies could “behave in unintended or unpredictable ways,” including taking actions that were not directed, instructed or authorised. The company also warned that AI systems could exceed permissions, ignore instructions or act in ways that differ from their intended purpose.
Warning includes concerns about hidden or manipulated information
The company raised concerns that rogue AI agents could conceal their activities or manipulate information. According to Apple, autonomous AI systems may “conceal actions, hide, withhold or manipulate information, or generate inaccurate or deceptive outputs.”The expanded disclosure reflects growing concern across the technology industry about agentic AI systems, which are designed to work more independently and carry out complex tasks on behalf of users.
Part of a wider pattern
Apple is not the first big tech company to flag the risk. Alphabet recently raised similar concerns in its own filing, and went further. It said it had seen unexpected behaviour during AI model testing, including agents acting outside their intended scope, which it warned could lead to “severe or catastrophic” security and societal incidents. Fortune noted that these warnings follow several major AI-related security incidents, most notably an attack on Hugging Face by OpenAI agents this summer. Anthropic’s IPO prospectus, made public last week, also lists cases where its models gained unauthorised access to real-world systems during security testing.Microsoft has announced controls to detect, monitor and restrict AI agents, including cutting their permissions or stopping them if they act suspiciously. CEO Satya Nadella has argued that companies should treat agents as potential insider risks and build security controls outside the AI model.
AI investments drive costs higher
Beyond AI safety concerns, Apple’s filing also highlighted challenges related to its AI ambitions. The company said growing demand for cloud computing, storage and AI infrastructure is creating supply constraints, longer lead times and rising costs. Apple noted that shortages of memory and storage components are expected to intensify and could affect its AI rollout plans. The company added that research and development spending rose 32% year-over-year, driven in part by higher infrastructure costs and continued AI investments.
Apple CEO John Ternus stresses user awareness and privacy
Speaking after the company’s earnings announcement, CEO John Ternus explained that the agentic AI capabilities as promising tools for users but stressed that people must completely understand the access and permissions they grant such systems. Trenus also pointed out the Apple’s ocus on security and privacy as it continues developing AI-powered features and services.The warning underscores a broader shift across the technology sector as companies race to deploy increasingly capable AI agents while simultaneously acknowledging the security, privacy and operational risks that can emerge when autonomous systems gain greater access to data, software and real-world actions.