Elon Musk’s platform, X, is testing a new way to fact-check posts by using its AI, Grok, to help write Community Notes. This new feature, called “Collaborative Notes,” aims to speed up the process of debunking viral misinformation, which often spreads faster than human contributors can react.
The process starts when a seasoned Community Note contributor asks for help on a post. Instead of waiting for a human to draft a response, the system triggers Grok to write an initial note. From there, human contributors take over. They can rate Grok’s draft, suggest better wording, or fix errors. X says the system will update the note in real-time as more feedback comes in. Eventually, the software decides if the updated version is good enough to show the public.
Keith Coleman, who leads Community Notes at X, says this isn’t the company’s first time using AI for this purpose. A previous pilot program allowed developers to build their own AI note-writers. One of those AI accounts has already written more than 1,000 notes that humans rated as “helpful.” However, this new experiment with Grok makes the process much more direct for regular contributors.
The main goal here is speed. Critics often point out that Community Notes take too long to appear, sometimes showing up days after a lie has already gone viral. By letting Grok handle the first draft, X hopes to get accurate information in front of users much quicker. Coleman also noted that this setup helps Grok learn. Every time a human corrects the AI or gives it a high rating, the model gets a little smarter.
Of course, letting an AI handle fact-checking is risky. Grok has a track record of making mistakes or “hallucinating” facts. To keep things under control, X is limiting the tool to “top writers” for now. These are the most trusted contributors on the platform. If the experiment goes well, X plans to let more people use it and might even share the technology with outside developers.
