AI to Write Community Notes for Fact Checking on X

X, the social media platform run by Elon Musk, has launched a pilot for an “AI Notes Writer,” a new API allowing developers to build AI systems that propose fact-checking notes on posts, with final judgment still in human hands.

The initiative builds on X’s existing Community Notes feature, where crowdsourced fact-checks are surfaced only if users from differing political perspectives rate them as helpful.

AI Note Writers follow the same rules: they must earn credibility through helpful contributions and cannot rate others’ notes. Their role is limited to proposing context, especially on posts flagged by users requesting notes.

“This has the potential to accelerate the speed and scale of Community Notes,” the company said on X, emphasising that humans will remain in control of what ultimately gets shown.

To participate, developers need to sign up for both the X API and the AI Note Writer API. Each AI Note Writer must pass an admission threshold based on feedback from an open-source evaluator trained on historical contributor data. Only notes from admitted AI writers can be surfaced to the broader community.

The company mentions that one can use GitHub actions and Grok or other third-party LLMs to build the AI Note Writer.

At launch, AI-written notes will be marked distinctly and held to the same transparency, quality, and fairness standards as human-written ones. The company also published a supporting research paper co-authored with academics from MIT and the University of Washington, outlining the approach’s potential and risks.

While the pilot begins with a small group, X says it plans to expand access gradually. The company hopes this experiment creates a feedback loop where AI models improve by learning from human judgment, without replacing it.

If successful, this could mark a turning point in how generative AI collaborates with people to reduce online misinformation at scale.