Claude Tag can now use a user's personal connections when that user invokes it in a Slack channel. Anthropic announced the update on September 24, rolling out to the Team plan with Enterprise expansion to follow.
You can therefore ask Claude in a channel to work with information already connected to your personal account, such as a Drive document or customer system data. Other participants do not thereby gain permission to invoke your personal connection.
Permissions are private; the answer may be shared
That distinction does not guarantee all information Claude produces remains private. If the answer is posted in the channel, channel members can read it. Users can therefore review the answer before publication or choose an automatic mode where the system may request review when it detects sensitive content.
According to Anthropic, Enterprise administrators will be able to require review before publication. As of the announcement, this does not mean all Enterprise capabilities are already available.
When a personal connection is not enough
The permissions update does not turn a personal connection into a shared tool the agent invokes without the user present. Anthropic clarifies that scheduled work and actions Claude initiates itself rely on connections an administrator has added to the channel.
This separates a one-off request to consult personal data from a regular team-serving process. Comparing a personal document with project status, for example, differs from monitoring an issue that needs to continue when the employee is offline.
In an active Claude Tag environment, the company says personal connections require no additional installation: the first time a request needs a connection, Claude asks for approval. The announcement included no dedicated availability for Israel or new Hebrew support. The important user decision is which information to allow Claude to read and what is appropriate to show in the channel.