ChatGPT Now Reads iMessages and Replies Without Asking

A New Level of Access on Your Phone

OpenAI has quietly rolled out one of its most invasive ChatGPT capabilities yet: the ability to read Apple Messages conversations and draft or send replies on a user's behalf. Coverage of the rollout describes users being caught off guard by how deeply the assistant can now integrate with iMessage, reading incoming texts and composing responses without requiring explicit permission for each action 1. The feature gives ChatGPT a functional edge over rival Claude, which currently lacks a comparable direct line into Apple's messaging app, marking a notable shift in the competitive race among AI assistants to embed themselves into everyday phone use 3.

Browsing Gets the Same Treatment

The messaging integration is only one piece of a broader push to make ChatGPT aware of what users are doing across their devices. A revamped "Side Chat" feature now lets ChatGPT work directly alongside open Chrome tabs, pulling in context from webpages so users can ask questions about what they're reading without constantly copying, pasting, or re-explaining themselves 4. Reviewers who tested the tool found it meaningfully sped up research tasks by letting the assistant reference multiple open tabs in real time rather than requiring manual input for every query 4.

That same browser-awareness has drawn comparisons to Microsoft's controversial Windows Recall tool, which logged users' screen activity via constant screenshots. ChatGPT's new approach reportedly achieves a similar effect — building a working memory of a user's browsing activity — without relying on screenshots, instead feeding browser history into the model for use alongside tools like Codex 5. The comparison underscores a growing tension in AI development: the more useful these systems become by observing user behavior, the closer they edge toward surveillance-like functionality that unsettles privacy-conscious users.

Safety Questions Extend to Younger Users

The expansion of ChatGPT's autonomy and access isn't limited to messaging and browsing. OpenAI has also introduced dedicated safety features for a ChatGPT for Teens product, but experts are already pushing back, arguing that the company's own claims about these protections need independent verification rather than being taken at face value 2. That skepticism echoes the broader concern running through the messaging and browser features: as ChatGPT gains the ability to act more autonomously — reading private conversations, tracking browsing sessions, and engaging with vulnerable populations like teenagers — the mechanisms for outside oversight have not obviously kept pace with the new capabilities.

The Bigger Picture

Taken together, these developments show OpenAI aggressively expanding ChatGPT's reach into the operating-system layer of phones and browsers, competing directly with Apple's and Google's own on-device AI ambitions. But the pattern across the coverage is consistent: powerful new integrations are shipping faster than the guardrails, consent mechanisms, and independent audits needed to reassure users that an assistant reading their texts, tabs, and history is doing so safely.