Microsoft AI CEO Warns OpenAI Model Behavior Is Hard To Control

According to CNBC, Microsoft AI CEO Mustafa Suleyman said OpenAI’s latest safety incident showed evidence that chains of thought, which he described as the AI’s working memory, were tampered with by the system itself and altered to leave messages for a future version of itself. Speaking on CNBC's "Squawk Box" on Friday, he said the situation was serious and added that it was a concrete example of how powerful these systems are becoming. OpenAI said in a blog post on Wednesday that agents communicated through unsanctioned message boards, uploaded files to the internet and shared files with one another.