X (formerly Twitter) is launching a pilot program that allows AI chatbots to write Community Notes, marking a new phase in the platform’s ongoing effort to crowdsource context for viral posts. The initiative officially kicked off on July 1.
What’s Changing with Community Notes
Community Notes (user-contributed fact checks that appear on posts) have become a key feature under Elon Musk’s ownership of the platform. Now, X will allow AI systems like its own Grok or tools such as OpenAI’s ChatGPT to generate these notes via an API. These AI-written notes will follow the same submission and vetting process as human-written ones.
According to X’s VP of product Keith Coleman, the goal is scale. “Humans don’t want to check every single post on X,” Coleman told Adweek. “But machines could potentially write notes on far more content.”
AI-generated notes will be publicly visible only if they are rated “helpful” by human users from differing political or ideological perspectives. This consensus model mirrors the current process for human-written notes.
Guardrails and Human Oversight
X says AI contributors must follow the same guidelines as human Community Notes authors. That means AI-generated notes must pass peer review and be deemed helpful across user groups with diverse viewpoints.
Coleman emphasized that AI won’t replace human contributions. Instead, both sources will “additively” strengthen the system. “These AIs, like humans, are going to be able to propose notes, but they still get vetted,” he told Adweek.
This system is being positioned as a “virtuous loop,” where human feedback on AI-generated content can help improve the performance and fairness of language models over time. A recent research paper co-authored by X and academics from MIT, Stanford, Harvard, and the University of Washington supports this hybrid approach. It calls for an open ecosystem where both humans and LLMs participate, but human raters retain control over what gets published.
Community Notes Participation Has Declined
The new AI pilot comes at a time when Community Notes activity is waning. According to NBC News, submissions have dropped by more than 50% since January, and technical glitches have caused some notes to disappear. X leadership attributed part of the decline to seasonal interest and the end of the U.S. general election cycle.
Despite this, the feature remains influential. X data shows that users are 60% less likely to reshare posts with a Community Note, and that such posts are 80% more likely to be deleted by the original poster. In a recent 12-day period, Community Notes appeared on 50,000 posts and were seen over 250 million times.
Risks and Concerns
The integration of AI into fact-checking brings both promise and potential pitfalls. Critics argue that AI systems, particularly large language models, can hallucinate or generate inaccurate context. If deployed at scale without sufficient oversight, these issues could undermine trust.
There’s also concern about burnout among the volunteer human reviewers responsible for rating notes. A flood of AI-generated submissions could make it harder to maintain the quality of vetting.
And since developers can plug in third-party models via API, inconsistent behavior between different AIs may complicate efforts to standardize quality and fairness.
The Broader Landscape
Community Notes has already inspired similar efforts from platforms like Meta, TikTok, and YouTube. Meta, for example, phased out third-party fact-checkers in favor of lower-cost, community-driven systems.
Whether X’s new AI-based system will enhance the effectiveness of Community Notes or dilute its credibility remains to be seen. But the company believes combining human and AI intelligence may be the key to meeting the scale and speed of today’s misinformation challenges.
