Apple Warns AI Agents Could Act Unpredictably

Apple's fiscal 2026 annual report says AI agents may act without human approval and may not follow instructions.

Updated on Nov. 4, 2026 in Artificial Intelligence

Apple Warns AI Agents Could Act Unpredictably

Live Poll

Do you trust AI tools to act on your behalf without close supervision?

Apple warned in its fiscal year 2026 annual report, filed with the U.S. Securities and Exchange Commission, that AI agents may act unpredictably. The company said such agents can execute code, transmit communications and complete transactions, sometimes without human instructions or approvals.

Why it matters

The filing flags operational and security risks from autonomous AI systems, including circumvention of safeguards and exploitation by malicious actors. Apple also said uncertainty around AI and superintelligence regulation and negative consumer perceptions could weigh on its business.

The warning appears in a formal SEC annual report, but the extraction provides no quantified risk metrics. It is not stated whether Apple has experienced any AI agent failures to date.

The players

Apple Inc.

Apple is the Cupertino, California technology company that makes the iPhone, Mac and related software and services.

Anthropic

Anthropic is an artificial intelligence company that develops the Claude family of AI models.

The details

Apple's report says AI agents may interact with internal and external systems and may be built by Apple or by third parties. Apple said it plans to add controls to Full Disk Access settings on Mac computers to prevent AI-agent risks and leaks of personal information.

Apple filed its fiscal year 2026 annual report with the SEC on November 3, 2026.
Anthropic filed a securities registration warning about advanced AI in October 2026.

Big Picture

Apple's risk factors fall under the SEC's annual-report disclosure rules, which require public companies to describe material risks in their Form 10-K filings. Its warning about AI agents extends that practice to autonomous software systems that can act without direct human approval.

Users of Apple Mac computers may see additional controls over Full Disk Access once the planned changes roll out. Consumers handling personal data through AI agents could face new privacy protections and permission prompts.

The takeaway

Autonomous AI tools can take actions without direct approval, so reviewing the permissions you grant software is worthwhile. Check your Mac's Full Disk Access settings periodically as Apple rolls out new controls.