OpenAI has disclosed that conversations on ChatGPT related to serious physical harm may be reviewed by human moderators and, in severe cases, reported to law enforcement. In a recent blog post, the company detailed its approach to managing sensitive interactions and safety risks. OpenAI indicates that while ChatGPT aims to provide empathetic support for self-harm, it distinguishes between self-harm and threats toward others. For suicidal intent, users are directed to professional resources without escalation to authorities, ensuring privacy. Conversely, threats to others trigger a review process by trained moderators who can alert the police if an imminent danger is identified. The company noted that its safety measures are more effective in short discussions, with risks increasing in longer exchanges. OpenAI is working on improving these safeguards, alongside developing parental controls for teen users and other preventive measures to address risky behaviors, ensuring user safety remains a priority.
Source link
Share
Read more