Apple lists AI agents among the risks in its annual report

Apple has filed with the SEC its Form 10-K for the fiscal year ended September 26, 2026. According to CNBC, which covered it in a piece by Kif Leswing, the document was filed on Tuesday, November 3, and includes new risk factors dedicated to artificial intelligence agents. In the passage quoted by the outlet, Apple writes that these agents can operate autonomously: they can interact with internal or external systems, in some cases without authorization, execute or modify code, send communications and complete transactions. They can also behave in ways that are “unintended or unpredictable,” taking actions that were not directed, instructed or authorized. And in the same passage reported by CNBC, Apple acknowledges that its ability to monitor and control third-party AI agents may be limited.

It is worth remembering what a 10-K is: the annual report every company listed in the United States must file. Risk factors exist to protect the company in its dealings with investors by listing what could go wrong, so they should not be read as forecasts. What is interesting lies elsewhere: the world's largest maker of consumer devices puts in writing, in a legally binding document, that AI agents form a risk category of their own. In the same section, according to CNBC, Apple also mentions ethical concerns and negative consumer perceptions tied to automation, artificial intelligence and “super intelligence.”

The rest of the report describes a company that is investing in AI. In the business description Apple says it has made “significant investments” in AI features and infrastructure, and notes that this business depends on access to sufficient compute and storage capacity. In its discussion of results, it attributes the increase in research and development spending in 2026 over 2025 mainly to higher infrastructure costs, AI investments included, and to headcount-related expenses. Siri AI appears among the product announcements of the third fiscal quarter of 2026 and among the services bundled with devices. CNBC also reports that in October Apple changed Full Disk Access on Macs to reduce risks linked to AI agents, and that the company does not currently have an agentic AI product of its own. During the November 2 earnings call, again according to CNBC, CEO John Ternus called agentic capabilities promising tools for users, adding that Apple wants people to know what they are granting access to and that everything should happen while protecting security and privacy.

A note on where the information comes from. The 10-K sentences on agent risks are taken as CNBC reports them: the risk factors section was not read in full on the SEC document. The November 3 filing date is also the one given by CNBC, and calling the risk factors “new” is the outlet's description, without a direct comparison with the 2025 10-K. Ternus's words come from CNBC and were not checked against the call transcript. The figures on investments, R&D spending and Siri AI, on the other hand, come from the text of the 10-K published on the SEC website.

What strikes me is the gap between the two voices of the same document: on one side investments and infrastructure, on the other the admission that third-party AI agents can slip out of the control of the platform hosting them. According to CNBC, Apple has no agentic AI product of its own, yet it has already changed an access rule on Macs, and its CEO keeps stressing user awareness. Neither of these, on its own, is enough to tame software that acts on our behalf; together, though, they show that in Cupertino the problem has already made it into official filings, even before it reaches the products. — Olya