What happened

The U.S. State Department’s “Catch and Revoke” programme uses AI tools focused on natural language processing (NLP) and keyword-based pattern recognition to analyse social media posts, news reports, and public data for phrases, hashtags, or imagery interpreted as endorsing Hamas or other designated terrorist groups.

Terms like “intifada” or “from the river to the sea” are flagged, alongside contextual sentiment assessments. 

However, critics highlight concerns about the tools' reliance on simplistic keyword matching and their potential inability to distinguish between nuanced political discourse (such as criticism of Israeli policies) and genuine terrorist sympathies.

Civil liberties advocates also warn the programme undermines First Amendment principles by outsourcing the policing of speech to potentially error-prone AI.

Why it happened

The Trump administration framed the policy as a national security necessity, citing the US Immigration and Nationality Act’s broad authority to revoke visas if holders “threaten public safety”. 

Officials claim past administrations neglected enforcement, emphasising the role of AI in analysing “publicly available information” at scale. 

The programme aligns with executive orders targeting anti-semitism and pro-Palestinian activism, leveraging efficiency of AI to surveil large populations.