Grok in Group Chats Arrives on X for Premium+

US subscribers can now bring the xAI assistant into group conversations, but adding the bot suspends end-to-end encryption.

X rolled out support for Grok in group chats on Thursday, October 1, 2026. The initial release is limited to Premium+ subscribers in the United States using XChat. The update lets room administrators introduce xAI’s assistant into group conversations, where it can read text, inspect uploaded images, and answer questions from participants.

How Grok in group chats works

Any room administrator can bring the bot into a thread. That happens either through the chat settings menu or by adding the @Grok account like an ordinary participant, according to X’s documentation and early user reports.

Once added, the assistant operates on two levels. It answers whenever a member explicitly mentions @Grok, and it can also interject on its own when it spots relevant conversational context, according to tests shared by independent researcher Andrew Curran.

Santiago, an engineer on the X team, outlined the main capabilities active at launch: answering questions, summarizing missed conversation history for returning members, creating shared reminders, and generating images or video clips on demand. Because the model is multimodal, it parses both standard chat messages and photos or screenshots posted to the group.

The privacy cost of group AI

XChat provides group conversations secured by end-to-end encryption using the Juicebox protocol. In standard protected chats, only the participants hold decryption keys, meaning server operators cannot read message contents passing through the network.

Inviting an AI assistant breaks that barrier. Support pages published by X confirm the operational reality: once @Grok joins a conversation, X and xAI gain direct access to every message sent while the bot is present. To parse chat flow, analyze shared images, and generate answers, xAI servers must process unencrypted text and media. If a group wants to restore full cryptographic secrecy, members must remove the bot.

What remains unconfirmed

Several operational details remain unannounced. The company has not provided a rollout schedule for users outside the United States, nor has it stated whether standard Premium tiers or free accounts will gain access later.

X’s documentation also leaves out which specific Grok model version powers the group feature. Furthermore, there is no public information on whether administrators can adjust the frequency of the bot’s unprompted replies, leaving open the question of how intrusive spontaneous interjections might be in high-volume rooms.

A quiet shift in group messaging

The rollout puts Grok on the same playing field as Meta AI across WhatsApp and Instagram groups. Yet the ability to speak up without a direct tag marks an important shift in behavior. The bot stops functioning merely as a command-line utility and starts acting like an active participant at the table.

For users, the trade-off is clear and practical. Having an assistant summarize missed chats and generate images brings convenience, but that utility comes at the expense of end-to-end message privacy.

Sources

• Official announcement from @chat: https://x.com/chat/status/2105743499807326599
• X Help Center (About Chat): https://help.x.com/en/using-x/about-chat
• X Help Center (About Grok): https://help.x.com/en/using-x/about-grok
• Santiago (@santiagomed), X team: https://x.com/santiagomed/status/2105745953647178195
• Andrew Curran (@AndrewCurran_), independent tester: https://x.com/AndrewCurran_/status/2105751619824652441 and https://x.com/AndrewCurran_/status/2105753310439641404
• Community testing report (@blankspeaker): https://x.com/blankspeaker/status/2103193095907279163