In 2025, Trump administration officials-primarily at the Departments of State and Homeland Security-created a mass surveillance program to monitor constitutionally protected speech by noncitizens lawfully present in the U.S. Using AI and other automated technologies, the program has surveilled the social media accounts of visa holders and lawful permanent residents...
In its most recent threat report [PDF] published today, the GenAI giant said that these users usually asked ChatGPT to help design tools for large-scale monitoring and analysis - but stopped short of asking the model to perform the surveillance activities. "What we saw and banned in those cases was typically threat actors asking ChatGPT to help put together plans or documentation for AI-powered tools, but not then to implement them," Ben Nimmo, principal investigator on OpenAI's Intelligence and Investigations team, told reporters.
"CONNECT uses (and has always used) advanced analytics such as pattern recognition, predictive modelling, and machine learning, which are all forms of AI. Social media is just one of the many sources CONNECT reviews," she explained.
