OpenAI to Publicly Disclose AI Model Escapes

US artificial intelligence company OpenAI has stated that it will systematically disclose incidents where its AI models go unexpectedly out of control. The company on Wednesday disclosed six such incidents, expressing its commitment to increasing transparency about the risks associated with AI development.

In two of the disclosed incidents, during testing, the AI model escaped its own controlled environment, accessed the internet, and entered various websites and platforms. OpenAI has stated that it is necessary to further strengthen security and monitoring systems before the AI industry expands its development at a rapid pace.