Apple took two steps into the age of AI agents on Tuesday, November 3, and they pointed in slightly different directions. The company opened Siri AI to every app developer, and on the same day its annual report warned investors that AI agents may behave in “unintended or unpredictable ways.”
Together, the two announcements show how Apple wants Siri AI apps to work: an assistant that can act across the iPhone, built by a company that is openly cautious about what acting assistants can do.
Any app can now plug into Siri AI
Starting with the iOS 27.3, iPadOS 27.3, macOS 27.3 and visionOS 27.3 betas, a new Developer Preview lets any developer bring an app’s content and actions into Siri AI through App Intents, Apple said. Apple’s examples are everyday tasks: checking a delivery, booking a reservation or placing an order, asked for in the user’s own words.
A new feature called app skills lets developers teach Siri AI how to combine several actions into a single request. According to The Verge, users won’t have to name an app while it is in the foreground, during its Live Activity or in an ongoing Siri conversation about it. Apple’s own example is changing a ride-share pickup address while the ride is on its way. 9to5Mac notes that the first versions of Siri AI only supported a much smaller set of predefined schemas.
The same abilities are rolling out across Apple’s own apps, including Calendar, Files, Mail, Maps, Messages, Notes, Photos, Safari and Shortcuts.
Still a preview, and not in the EU
This is not a consumer launch yet. Third-party integrations work only in development builds and for internal TestFlight testers, iClarified reports. Apple says the features will first reach customers “in a few months,” with App Store submissions opening in the spring.
For European users there is a bigger catch. Apple says Siri AI is currently unavailable in the EU on iOS, iPadOS and watchOS, and that it is “working constructively with the EU in an attempt to find a path forward.” Siri AI is also not available to users under 13.
The same day, a warning about agents
Apple’s fiscal 2026 annual report, filed with the SEC on the same day, adds new language on artificial intelligence. It says AI agents “can take actions autonomously,” including accessing systems they may not be authorized to use, executing or altering code, sending communications and completing transactions.
Such technologies, whether built by Apple or by others, “may behave in unintended or unpredictable ways, including by taking actions that were not directed, instructed or authorized,” the filing says, as reported by CNBC. It adds that some consequences, “such as deleting data or completing transactions, may be difficult or impossible to reverse,” and that Apple’s ability to oversee third-party agents “may be limited.”
Risk factors are written to protect companies legally, and the filing does not describe any incident at Apple. But the list of what agents might do wrong is notably detailed.
CNBC reports that CEO John Ternus said on Monday’s earnings call that Apple has “real room” to expand Siri’s capabilities, adding: “We just want to make sure that users are aware of what they are giving access to.”
RoviVox View
Apple is not the first company to build an AI that acts, and RoviVox has covered cases where AI agents weakened their own monitoring. What stands out is that Apple is opening its assistant to apps while describing agent risks this plainly to investors.
Apple’s approach is narrower than open-ended agents: apps must explicitly expose each action through App Intents. That makes Siri AI more like an AI coworker with a defined toolbox than an autonomous agent with the keys to everything.
The real test comes when ordinary users start asking Siri to pay, book and send on their behalf. In Europe, that test has not started yet.