X (formerly Twitter) is now allowing developers to design AI “Note Writer” bots that generate Community Notes—contextual fact-check summaries—which can be submitted when users request additional context on a post.
These AI-generated notes are only published if a diverse group of human contributors find them helpful, and every note will be clearly labeled as crafted by an AI.
To earn the ability to write notes, these AI bots must undergo tests and receive ongoing approval based on the usefulness of their contributions.
X’s vice president of product, Keith Coleman, emphasizes a “test mode” rollout, stating that humans will remain the ultimate judges of whether a note is useful enough to display publicly.
This pilot initiative aims to scale Community Notes by enabling quicker and broader generation of note content, leveraging AI to increase reach while maintaining human oversight.
Still, experts warn that AI-written notes may present factual errors or misleading statements—even when presented confidently—and could burden human reviewers.
X argues that by crowdsourcing ratings on AI outputs, it will improve AI accuracy over time, creating a feedback loop that refines both content quality and bias control .
Nevertheless, researchers caution that reliance on AI could cause note overload for human moderators, and there’s concern AI might eventually assume rating roles—risking the integrity of the user-driven moderation process .
Despite these concerns, some see merit in the approach—particularly in handling repetitive misinformation quickly—while others question whether the system’s volunteer base can scale to manage the surge in content .
