OpenAI is enhancing its safety protocols following the tragic Tumbler Ridge, B.C., mass shooting, where shooter Jesse Van Rootselaar had a ChatGPT account banned months prior. Ann O’Leary, OpenAI’s VP of Global Policy, informed Canada’s AI Minister, Evan Solomon, that the company has partnered with mental health and law enforcement experts to improve referral criteria for potentially dangerous accounts. Despite banning Van Rootselaar’s account, OpenAI did not initially report it to authorities, as it did not meet their threshold for imminent threat. However, under new protocols, they would have referred the issue if discovered today. Following the incident, OpenAI identified a second account belonging to the shooter and shared it with police. Solomon expressed disappointment over OpenAI’s lack of substantial solutions during meetings, but the company aims to establish better communication with law enforcement and enhance detection systems for policy violators. The government is currently reviewing OpenAI’s commitments.
Source link
