OpenAI Launches Apple Messages Plugin, Letting ChatGPT Read, Draft, and Send Texts
OpenAI has launched an Apple Messages plugin that lets ChatGPT read, sort, draft, and send texts on a user's behalf. The company says the plugin runs locally and does not index all messages, though it has not detailed exactly how that works.
OpenAI has released an Apple Messages plugin for ChatGPT that allows users to connect their Messages inbox to the chatbot, enabling it to sort, analyze, edit, draft, and send texts on their behalf.
The plugin, announced August 20, 2026, also integrates with Codex and ChatGPT Work, extending the feature beyond personal use into professional workflows. A promotional video from OpenAI demonstrates a user asking ChatGPT to suggest follow-up messages to contacts based on the previous day's conversations. The plugin can also delete messages, draft and send replies as the user, and search through message history for specific information.
How it works
According to OpenAI, the plugin runs locally on a user's machine and does not create a full index of all their messages. The company has not provided technical specifics on how message data is processed, stored, or excluded from indexing. TechCrunch has requested additional detail from OpenAI and has not yet received a response.
On message sending specifically, OpenAI is advising caution. The company recommends users actively monitor what ChatGPT sends on their behalf and discourages enabling persistent approval settings. Turning on persistent approval, OpenAI warns, "removes your final chance to review a message before ChatGPT sends it as you."
Pricing and availability
OpenAI has not disclosed pricing details specific to the plugin. Availability details — including whether it requires a paid ChatGPT tier or works with the free version — were not specified in OpenAI's announcement.
What this means
Giving a chatbot direct read/write access to a user's private text messages is a significant expansion of ChatGPT's reach into personal data, and it raises the stakes on questions OpenAI has not yet fully answered. The claim that the plugin "doesn't create an index of all someone's messages" is doing a lot of work here — it's unclear whether messages are processed transiently, cached temporarily, or handled some other way, and OpenAI has offered no technical documentation to clarify.
The integration with ChatGPT Work signals OpenAI's ambition to position this as an enterprise productivity feature, not just a consumer convenience. That raises additional questions about compliance, data retention policies, and whether business messages sent through the plugin are subject to different handling than personal ones.
The explicit warning against persistent approval is notable — it suggests OpenAI itself is wary of the failure modes here. An AI system autonomously sending text messages as a user, without per-message review, opens the door to embarrassing, incorrect, or damaging texts sent without the user's knowledge. Until OpenAI publishes more detail on data handling and provides independent verification of its local-processing claims, users connecting their message history should treat this as an early-stage integration with real privacy tradeoffs, not a fully audited feature.
Related Articles
GPT-6 Astra Ultrafast Claims Up to 8x Faster Token Generation on NVIDIA Blackwell GPUs
GPT-6 Astra Ultrafast, running on NVIDIA Blackwell GPUs, is available now in the OpenAI API and to eligible ChatGPT Work and Codex users. NVIDIA says it generates tokens up to 8x faster than Astra Standard mode. Pricing and context window details were not disclosed in the source.
ChatGPT adds virtual try-on and Favorites in global shopping update built on Images 2.5
OpenAI launched two shopping features in ChatGPT globally on Thursday: a virtual try-on tool for clothing and accessories, and a Favorites function for saving products. OpenAI says both use its newly launched ChatGPT Images 2.5 model. Pricing and plan availability were not disclosed in the source reporting.
OpenAI publishes startup guide for GPT-6 family covering model choice, reasoning effort and tool coordination
OpenAI has published "A model guide for the GPT-6 family," a practical guide aimed at startups. It covers choosing GPT-6 models, tuning reasoning effort, improving prompts and skills, coordinating tools, and preparing workflows for production. The summary gives no pricing, context window or benchmark figures.
OpenAI publishes startup guide to choosing and deploying GPT-6 models, with reasoning-effort tuning
OpenAI has published a practical guide for startups building on the GPT-6 family. It covers model selection, reasoning effort, prompts and skills, tool coordination, and production workflows. The available summary discloses no pricing, context window, or benchmark figures.
Comments
Loading...