US4 min(s) read
Published 13:49 26 Jan 2026 GMT
A widely shared video from Minneapolis showing federal agents fatally shooting ICU nurse Alex Pretti captured national attention and intense public debate.
Pretti, a 37‑year‑old US citizen and nurse at the Minneapolis Veterans Affairs Health Care System, was killed by federal immigration agents during an enforcement action on January 24, 2026.
Videos circulating online showed a chaotic encounter in which he was tackled to the ground before being shot. Critics have argued those videos challenge official accounts that he posed a clear threat to agents.
Protests have erupted locally and in cities across the US, prompting calls for thorough investigations into federal immigration enforcement tactics.
A YouTube creator known as I Ask AI took this controversy one step further by uploading a video titled “I Sent AI the ICE Shooting Video From Minneapolis and Let It React” on Sunday.
In that video, the creator fed footage of the encounter to an AI tool after first briefing it on the situation so the model could interpret what it saw in the raw video rather than simply echo official narratives.
The experiment also included showing the AI a purported letter from former Florida Attorney General Pam Bondi related to ICE activity. The creator said the goal was to see whether the AI could offer insight “without immediately collapsing into talking points.”
The creator prefaced the experiment with a disclaimer encouraging viewers not to treat the AI as an expert or final authority, but rather as a test of whether such tools can handle complex, real‑world situations.
The experiment included not just the police‑seen clips, but also questions about practical safety advice for everyday people during enforcement actions.
When the AI evaluated the Minneapolis footage on its own terms, it described a federal immigration enforcement operation that “turns lethal almost immediately” and emphasised that how the situation unfolded mattered more than official explanations.
The model noted agents arrived in tactical gear and “close distance aggressively in a public space,” arguing that this posture suggested force was the priority.
The AI said the visual record indicated escalation initiated by federal agents and that “there are moments that strongly suggest he is disarmed before the shots are fired,” adding that once agents had physical control on the ground, continued use of lethal force was avoidable.
In its summary, the AI said the video “does not come across as controlled or inevitable.”
On the question of whether Pretti posed a clear threat at the moment of the shooting, the AI asserted that if justification were present, the visuals would have shown an “unmistakable immediate threat.”
Instead, it noted confusion, pepper spray, a disarmed man, and gunfire. When the AI offered advice on how regular people should behave during such operations, it suggested avoiding proximity, leaving the area if possible, keeping hands visible, and refraining from sudden movements or arguments with agents in the moment, noting these steps could help people stay safe and live to resolve disputes later.
The video experiment also included showing the AI a screenshot of what was described as a letter allegedly from Pam Bondi, in which she tells Minnesota officials that ICE would withdraw if the state turned over its voter database.
When interpreting that document, the AI described such a tactic, if genuine, as coercive, using enforcement power to pressurise a state government over sensitive data.
It characterised the alleged letter’s content as turning immigration enforcement into a political instrument rather than a law‑enforcement function, and suggested that “mixing immigration enforcement with election pressure is reckless and corrosive.”
Responses to the AI’s take on both the video and the letter were part of the creator’s broader attempt to foster discussion about how people and technology intersect when confronting controversial events that have stirred widespread protests and demands for accountability.
.webp_SmgFsc?tr=cr-292.0.1244.952,f-webp,q-70,w-100)
