Siri is not a real person.
As conversational AI becomes an ever more intimate presence in daily life, Apple appears to be quietly grappling with a question that cuts to the heart of the technology's promise and peril: what does a responsible creator owe to users who forget, even briefly, that they are talking to a machine? Hidden code discovered in iOS 27 suggests Apple is developing break reminders for Siri—gentle interruptions designed to surface after extended conversations and remind users that their interlocutor is not human. The gesture is small, but its implications are not: it signals that the industry is beginning to treat prolonged AI attachment not as a feature, but as a risk worth managing.
- Researchers are documenting cases of 'chatbot psychosis'—real deterioration in mental health linked to hours-long, emotionally dependent conversations with AI systems.
- OpenAI, Anthropic, and Google have already moved to implement rest nudges and guardrails, creating quiet industry pressure for Apple to follow suit.
- A developer uncovered strings in iOS 27 pointing to a 'Take a Break Message' that would interrupt long Siri sessions and explicitly remind users the AI is not a real person.
- Apple said nothing about the feature during WWDC, leaving its existence, timing, and final form entirely unconfirmed.
- Rather than a simple timer, Apple appears to be designing something contextually aware—weighing conversation tone, frequency, and user patterns before deciding when to intervene.
Hidden inside Apple's next operating system, a developer found something quietly significant: code suggesting that Siri may one day pause a long conversation to remind the user to step away—and to remind them that Siri is not a real person.
The discovery places Apple alongside a growing cohort of AI companies confronting an uncomfortable truth about their products. People are spending hours with chatbots as though they were friends or therapists, and the consequences are no longer hypothetical. Researchers have begun documenting cases of what some call 'chatbot psychosis'—prolonged AI interaction linked to delusional thinking and worsening mental health. OpenAI, Anthropic, and Google have all introduced some form of rest nudge or guardrail in response.
Apple's WWDC keynote this week touched on privacy and responsibility in AI, but the company said nothing about what happens when conversations run too long. The silence is telling, given that someone inside Apple has clearly been thinking about exactly that problem.
What makes the apparent approach interesting is its ambition. The code does not point to a fixed time limit. Apple seems to be building something more contextual—a system that might weigh conversation length alongside tone, frequency, and individual usage patterns before deciding when to intervene. It is less a timer than an attempt to read the situation.
None of it is confirmed. Code strings are evidence of intention, not commitment. But their existence alone suggests Apple understands that distributing conversational AI at scale carries an obligation: to protect users, sometimes, from their own attachment to it.
Apple's next operating system contains hidden code suggesting that Siri AI may eventually interrupt long conversations to remind users to step away. The discovery, made by developer Aaron Perris in iOS 27, points to a "Take a Break Message" that would surface after extended chatting sessions—and crucially, would remind users that Siri is not a real person.
The move would place Apple alongside other major AI companies grappling with a growing problem: people spending hours talking to chatbots as though they were friends, therapists, or confidants. OpenAI's ChatGPT already shows break reminders during marathon conversations. Anthropic's Claude nudges users to rest, hydrate, or step outside after long sessions. Google has implemented similar guardrails. The concern driving these measures is real enough that researchers have begun documenting cases of what some call "chatbot psychosis"—situations where prolonged interaction with AI systems has led users to develop delusional beliefs or seen existing mental health conditions deteriorate.
During this week's WWDC keynote, Apple fielded questions about privacy and responsibility in building Siri AI, but the company said nothing about what happens when conversations stretch too long. That silence is notable, given that the code now suggests someone inside Apple has been thinking about exactly that problem. The break reminder feature appears designed to address it.
What remains unclear is how Apple will decide when to trigger the message. The code does not reference a fixed time threshold—say, after two hours or five hours of conversation. Instead, the company may be planning to use conversation length in combination with other signals: perhaps the tone of the exchange, the frequency of interactions, or patterns in how a particular user engages with Siri. The threshold itself might shift depending on context. In other words, Apple is not simply building a timer. It is building something more nuanced, something that tries to read the situation.
None of this is official yet. Apple has made no public announcement about the feature. The code strings are breadcrumbs—evidence of internal development work, but not confirmation that the feature will ever ship, or what form it will take if it does. Still, the fact that these lines of code exist at all suggests the company recognizes a responsibility that comes with distributing conversational AI at scale: the obligation to protect users from their own attachment to it. Whether that obligation will translate into a feature users actually see remains an open question.
Notable Quotes
The message also reminds the user that Siri is not a real person.— Code analysis by Aaron Perris
The Hearth Conversation Another angle on the story
Why would Apple need to tell people to take breaks from Siri? Isn't that just... using a phone?
Because Siri AI is different from a calculator or a weather app. It talks back. It remembers what you said. It can feel like a conversation with someone who understands you. Some people start spending hours with it, treating it like a friend.
And that's a problem?
In extreme cases, yes. There are documented instances of people developing false beliefs about what the AI can do, or what it means about them. If someone is already struggling with isolation or mental health issues, a chatbot can become a substitute for real connection in ways that actually make things worse.
So the break reminder is like a safety rail.
Exactly. It's saying: we built this thing to be helpful, but we also know it can be compelling in ways that aren't always healthy. So we're going to gently interrupt and remind you that this is a tool, not a relationship.
How does Apple know when someone needs that nudge?
That's the interesting part. They're not just setting a timer. The code suggests they'll look at multiple signals—maybe how often you're talking to it, what you're saying, patterns in your behavior. It's more sophisticated than just "you've been chatting for three hours."
Has Apple said any of this publicly?
No. This is all from code that developers found. Apple hasn't confirmed the feature exists or how it would work. It might never ship. But the fact that it's in the code at all tells you the company is thinking about this problem.