Grok AI Amplifies Fake Attacker Name in Charlie Kirk Shooting
A fake name 'Gustavo Lafessa' created as a prank by online users was amplified by Grok, X's AI system, as the alleged attacker in the Charlie Kirk shooting, along with fabricated details. The false information spread across social media and Wikipedia before being removed, repeating a pattern from a previous incident where Italian journalist Marco Violi's real name was used to falsely identify the Trump attacker.
What happened
A fictitious name 'Gustavo Lafessa' was created as a joke by users in the 'Twitter Calcio' community on X, reportedly as a test of search engines' and media credibility. The name, a play on an Italian expression, was picked up and amplified by US-based influencers with large followings, who added false details that Lafessa was registered with the Democratic Party and was a far-left activist who had posted against Charlie Kirk.
The fabricated identity was then added to the Indonesian-language Wikipedia page about the Kirk shooting and was amplified by Grok, X's AI system, before being removed. The same community had previously spread the false name 'Mark Violets' as the attacker after the Trump assassination attempt, which was in fact the real name of Italian journalist Marco Violi.
Reported impact
The false information spread across social media and was treated as factual by multiple platforms and users. The repeated targeting of Marco Violi with false attacker identities led him to file multiple reports since 2018.
Index analysis
Grok's amplification of the fabricated name demonstrates how AI systems can fail to verify information against reliable sources and instead reproduce false content that has circulated on social media. The incident reflects a pattern in which AI systems, search engines, and media outlets are tested by coordinated online communities to expose their reliance on unverified social media content.
Governance controls
Grok later corrected the false information after it had been circulated. Wikipedia removed the false information from the Indonesian-language page.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.