OpenAI Strengthens AI Security Measures to Prevent Model "Out of Control," Latest Model Training Paused for Two Weeks

According to reports from Financial界, following recent cybersecurity incidents that have raised concerns about the potential for AI tools to go out of control, OpenAI announced that it will adopt stricter systems to monitor and safeguard the artificial intelligence models under development. OpenAI stated that it plans to further track the behavior of its most capable, yet-to-be-released models when solving problems and using various online tools, with the goal of alerting the security team within 30 minutes upon detecting concerning behavior. OpenAI also revealed that the company has added control measures to prevent certain AI models from connecting to the internet to perform higher-risk tasks. Additionally, OpenAI will require stricter isolation measures, known as "sandboxing," during the training and evaluation of AI models.