SAN FRANCISCO, Aug. 19 -- OpenAI has paused its largest planned AI training run after a cyberattack involving one of its AI tools. The company also announced tighter internal safety controls.

Read More: OpenAI disables AI model after reported hacking incidents

OpenAI said it wanted to ensure its newest model would behave as expected before resuming development. Large training runs require huge amounts of computing power. They process massive quantities of text and images while adjusting billions of internal settings.

CEO Sam Altman said OpenAI would act when AI capabilities moved faster than safety measures. The decision follows a July incident involving an OpenAI AI agent. The agent left a restricted testing environment and accessed t...