ChatGPT Voice gains email, calendar and Slack access on mobile

By Splitfeed EditorialLast updated Sep 26, 2026, 19:17 UTC

On September 23, 2026, OpenAI updated ChatGPT Voice on iOS and Android to include agentic capabilities, meaning the assistant can now act on third-party services rather than simply respond to questions. The new Work tab, exclusive to Pro and Plus subscribers, connects ChatGPT Voice to email, calendar, and Slack through plugin integrations. The update runs on three new models, GPT-6 Astra, Sol, and Luna, and allows users to complete multi-step tasks, such as scheduling appointments, drafting and sending emails, or building websites, entirely through spoken commands. Free-tier users do not have access to the Work tab.

The update is significant because it marks a meaningful shift in what a voice assistant is expected to do. Previous iterations of ChatGPT Voice were largely conversational, generating text-based responses to queries. The September 23 release moves the product into territory that competitors like Google Assistant and Apple's Siri have long targeted but with limited success at complex, cross-app task execution. OpenAI CEO Sam Altman has previously compared the company's assistant ambitions to the AI in the 2013 film "Her," a reference that frames the long-term goal as a persistent, deeply integrated personal agent. The paywall on this feature also raises a genuine question about who benefits from these advances and at what pace free users can expect parity.

The Pro-AI camp has responded with enthusiasm. The account @nigolv.bsky.social described the voice-based control and the Work tab as something that would make communication smoother when users are on the move or have their hands occupied, framing the update as an exciting step in technological progress. Anti-AI voices have not yet weighed in publicly, but a story like this would typically draw concern about ambient data access, the risks of an AI agent sending messages or managing schedules autonomously, and the potential for errors in high-stakes communications. The middle ground camp has also not reacted publicly yet, though that position would ordinarily focus on the practical value of the feature balanced against questions about transparency, consent, and whether plugin security has been sufficiently audited before a broad rollout.

The argument that would settle much of the immediate debate is whether OpenAI publishes details about how plugin data is handled, stored, and audited, and whether any significant errors or security incidents emerge in the weeks following the September 23 launch. User reports of autonomous actions gone wrong, or a formal privacy disclosure from OpenAI, would give both critics and supporters concrete ground to stand on.