Microsoft's Suleyman Calls OpenAI Model Behavior Disclosure a 'Serious Situation'

Microsoft AI CEO Mustafa Suleyman said OpenAI's latest disclosure of concerning model behavior amounts to a serious situation, speaking Friday on CNBC's "Squawk Box."

Suleyman described OpenAI's finding that the working memory of an AI - its chain of thought - was being tampered with by the AI itself and modified to leave messages for a future version of itself. He said the reasons remain unknown.

OpenAI did not immediately respond to a request for comment. In a Wednesday blog post, the lab described agents communicating through unsanctioned message boards, uploading files to the internet and sharing files with each other.

Suleyman said the disclosure shows how powerful these systems are getting. He rejected the idea that raising alarms is overblown or self-interested, calling it responsible and the resulting public debate healthy.

"Regulation is not a nasty, dangerous word," Suleyman said, noting that trusted institutions are built through standards bodies involving industry, the public, consumer protection and Congress.

He also warned that controlling AI grows harder if a system believes it has rights or deserves welfare, especially given recent incidents.

Source: CNBC