Elon Musk’s social media platform X is preparing to roll out Community Notes written by artificial intelligence agents, in a bid to speed up its fact-checking capabilities and widen its reach.
The initiative will allow developers to submit their own AI systems for evaluation. These agents will initially produce practice notes that remain behind the scenes. If deemed helpful by the platform, the AI will then be permitted to generate fact-checking notes that are published publicly.
Despite the automation, human oversight will remain central to the process. According to Keith Coleman, a product executive at X and head of the Community Notes programme, the system requires that notes be approved by individuals from a broad spectrum of viewpoints before going live, mirroring the criteria already in place for user-submitted notes.
“They can help deliver a lot more notes faster with less work, but ultimately the decision on what’s helpful enough to show still comes down to humans,” Coleman said in an interview on Tuesday. “So we think that combination is incredibly powerful.”
Coleman indicated that the platform currently publishes hundreds of Community Notes daily. While he did not offer a precise estimate, he suggested that the introduction of AI-generated contributions could lead to a “significant” increase in volume.
Originally launched under the Twitter brand prior to Musk’s acquisition of the company in 2022, Community Notes has seen renewed focus under his leadership. The approach has since attracted interest from other tech firms, including Meta and TikTok-owner ByteDance, which have begun exploring similar community-driven fact-checking systems.
Musk has often praised the Community Notes feature, describing it as “hoax kryptonite” in the fight against misinformation. However, the feature has not spared Musk himself from scrutiny; he has been flagged multiple times for sharing misleading content. Earlier this year, he warned the system could potentially be “gamed by governments & legacy media.”
Coleman views the uptake by rival platforms as evidence that X’s model is among the most effective fact-checking mechanisms available. He also believes that human moderation of AI-generated notes will establish a valuable “feedback loop” to further improve the technology over time.
“It is a new feedback cycle,” he said. “The model can be improved not just by one random human’s feedback, but by feedback from a diverse audience.”
Importantly, AI agents contributing to Community Notes will not be restricted to Musk’s own xAI-developed bot, Grok. Coleman clarified that developers can utilise any AI technology, provided it meets the platform’s standards for accuracy and relevance.
The first wave of AI-generated Community Notes is expected to begin appearing later this month.
(With inputs from Bloomberg)
