DOJ Uses Pangram AI Detector Against Defense in Texas Vote Case
TL;DR
- Federal prosecutors in Texas submitted Pangram AI-detection results as court exhibits and told Judge Mark Pittman the defense had 'turned into defense by ChatGPT.'
- Judge Pittman said he was 'disturbed' by filings that appeared 'clearly generated by artificial intelligence' in United States v. Khalwinder Bhengura.
- Defense attorneys Nicole Knox and Mario Herrera attributed AI drafting to a medical emergency that left Herrera four days to prepare the federal trial.
Federal prosecutors in the Northern District of Texas ran their opposing counsel's briefs through a commercial AI detector called Pangram and told the court that, in their words, 'virtually nothing of substance in these filings seems to have been drafted by a human.'
The case is United States v. Khalwinder Bhengura, where a 70-year-old Indian permanent resident was convicted of illegally registering to vote and voting in the 2024 election after falsely stating citizenship on a driver's license application. 404 Media reported that the government's filing told Judge Mark Pittman the matter 'has turned into defense by ChatGPT,' dismissed the opposing briefs as 'AI slop,' and argued the government should not have to engage substantively with arguments it believes a model produced.
Pittman said he was 'disturbed by counsel's repeated filings which subtly bend the law' through 'arguments so clearly generated by artificial intelligence.'
Defense attorneys Nicole Knox and Mario Herrera do not deny using AI. They say Knox had a medical emergency during trial prep, which left Herrera four days to ready a federal jury trial on his own. 'I represented my client to the best of my abilities within this limited timeframe. I utilized all available legal tools,' Herrera told the court, adding that 'The U.S. Attorney's Office did not identify any false citations.' Knox was blunter about what Pangram can and cannot prove. 'AI could have never come up with our argument,' she said, and pointed to the Northern District's disclosure rule, which, as she characterized it, requires transparency even if 'you obtain one sentence or phrase through the use of A.I.'
Pangram CEO Max Spero told 404 Media that law firms mostly use the tool internally, to monitor 'junior attorneys and paralegals,' not as an exhibit against opposing counsel. Our AI-detection tracker has logged 26 prior stories in the last 90 days, most involving classifiers aimed at strangers rather than the lawyer across the aisle.
Shared on Bluesky by 2 AI experts
Originally reported by 404media.co
Read the original article →Original headline: DOJ Submits Pangram AI-Detector Results as Exhibits to Argue Defense Briefs Were Written by ChatGPT