X is deploying an upgraded version of its Grok AI model to catch creators who profit off stolen content, redirecting monetised impressions and the money that comes with them, back to the people who originally posted it. The company said its latest Grok model can now detect duplicated content at three times the rate of its previous version, and has already flagged 1.5 million stolen posts in its most recent review cycle.

How the Detection Works

According to X's head of product Nikita Bier, the new system catches reposted videos and copied text posts even when users try to disguise them with watermarks, added introductions, or other edits. Instead of crediting engagement to the account that reposted the content, monetised impressions will now automatically flow to the original uploader. The crackdown extends to duplicated viral text posts too, Bier noted that one of the most commonly copied lines on the platform is a variation of "Twitter is like the smoking section of the internet."

As a direct result of these changes, X says more than $1 million in creator payouts will be redirected to original content owners instead of those who copied it. Nearly 4,000 accounts have already been removed from the platform's revenue-sharing programme as part of the sweep, and X has said repeated or intentional attempts to circumvent detection will result in permanent removal from the programme.

Alongside tackling copied content, X is also using Grok to detect engagement bait, such as posts promising to follow back users in exchange for replies. According to Nikita Bier, accounts repeatedly using these tactics may be removed from the platform's revenue-sharing programme and could face suspension.

The update is part of X's broader AI moderation strategy. The company has previously deployed Grok to identify bot accounts and other forms of platform abuse, expanding its use of AI beyond content generation to enforcement and moderation.

The update is expected to help original creators retain more of their earnings by reducing revenue lost to copied content, while making it harder for accounts that rely on reposting to monetise. It also reflects a broader trend of social platforms using AI to detect content misuse and strengthen creator protection.