X introduces AI-assisted Collaborative Notes

X has introduced a new experimental feature, Collaborative Notes, that allows community members to work with AI to add context to posts through a shared editing process.

According to the platform, Collaborative Notes are created when contributors request a note on a post. An AI system generates an initial draft, which is then refined through ratings and suggestions from other contributors. Notes can be updated over time as new feedback is received.

The platform noted, “When you request a note, AI drafts one - then the community refines it together through ratings and suggestions.”

The feature builds on the platform’s existing Community Notes system, which allows users to add context to posts that may be misleading or lack information. Collaborative Notes are written and updated by AI, using community input, and represent “a new approach to the public working together with AI.”

At launch, only requests from Top Writers on English-language posts will generate Collaborative Notes. Users must have a Writing Impact score of at least 2 to suggest improvements. It plans to expand access over time and adjust the feature based on feedback.

“This is a new, experimental feature, and we expect it to evolve based on contributor feedback,” X said.

Contributors can interact with Collaborative Notes by rating them, similar to existing Community Notes. Ratings help determine whether a note is marked as Helpful and displayed on a post.

Contributors can also suggest changes while submitting a rating. Suggestions can be edited or removed by changing the associated rating.

Collaborative Notes may be updated as new ratings and suggestions are reviewed. The system evaluates whether incoming feedback leads to a meaningful improvement before publishing a revised version.

Contributors are notified if a note they commented on is updated and whether their suggestion was used.

Each note includes a revision history that allows users to view earlier versions.

Collaborative Notes follow the same transparency principles as Community Notes. The code used to generate notes, including prompts and logic, is open-source. Notes and suggestions are also included in public data downloads.

It is exploring the introduction of ‘Collaborative Impact’ points, which would sit alongside existing Writing and Rating Impact scores and show contributors the effect of their suggestions.