TL;DR

Texas governor Greg Abbott, state attorney general Ken Paxton and New York representative Mike Lawler all shared or “liked” an AI-generated image purporting to show a downed US airman rescued by special forces in Iran over Easter weekend. The picture, originally posted by a pro-Trump X account, has been reshared more than 21,000 times and now carries a platform warning that it is “probable AI generated”.

Senior Officials Caught in a Familiar Trap

Abbott told his 1.4 million followers “this is so awesome” before deleting the post, adding “God is sending a message to our enemies!” Lawler captioned his repost “God Bless America!” The pattern is by now well established: a synthetic image attaches itself to a real, fast-moving news event — in this case, a rescue mission Donald Trump was due to discuss the same afternoon — and slips past elected officials before fact-checkers catch up.

This is not Abbott’s first encounter with AI fakery. Just last month he posted what he believed was footage of an Iranian plane being shot down by a US warship; the clip turned out to be from the World War II video game War Thunder. NewsGuard’s misinformation editor Sofia Rubinson, quoted by the Guardian, has previously warned that the most persuasive fakes “do not drastically distort the facts on the ground” — they fill in plausible detail in stories where verified imagery is scarce.

A Bipartisan Problem

The Guardian notes that AI-generated political fakes are not the preserve of one side. Democratic strategist Keith Edwards posted an AI image of Trump using a walker that drew 13.5 million views in December, and California governor Gavin Newsom routinely uses AI-generated images for political messaging. Hany Farid of UC Berkeley’s digital forensics group warns that in volatile situations such errors can be “dangerous”, adding noise to “an already complicated and difficult situation”.

Looking Forward

For UK readers, the episode is a useful proxy for what is coming. British platforms operate under the same incentive structures as X, the same generative tools are available, and the next general election cycle will land in a media environment where viral AI imagery is now routine. The case for media-literacy interventions, platform-side provenance signalling and senior-official training is no longer theoretical — and Whitehall is yet to decide which department actually owns it.