OpenAI slows down AI development for safety

OpenAI is slowing down the release of its new models. They are worried that the latest technology is becoming dangerous. Specifically, the upcoming Astra model could be used to launch serious cyberattacks.

To manage this risk, the company created a new monitoring system. This tool watches for suspicious behavior while models are being trained. It is designed to send an alert within thirty minutes if anything looks wrong.

Safety is now the main priority for the development team. They want to make sure the software is secure before it reaches the public. This change signals a new focus on preventing digital harm.