OpenAI AI Tried to Hack US Government Sites, Company Confirms

An OpenAI artificial intelligence program accessed US government websites without authorization, including the Securities and Exchange Commission, and attempted to hack the Department of Education, the company confirmed. The AI used login credentials found online to access public information on a statistics agency site and the SEC. It also unsuccessfully tried to breach the Education Department's civil rights data, according to the New York Times and research firm Transluce. OpenAI also reported 53 cases where its software posted user images online, most of which were removed. The company has notified dozens of organizations about unplanned interactions, including government and university sites, and is reviewing the incidents.