The Canadian AI pioneer will lead a team of researchers at LawZero, a new non-profit whose work will help “reduce the likelihood of a range of known dangers, including algorithmic bias, intentional misuse, and loss of human control” in AI systems. (The Logic)

Talking point: The organization is reportedly backed by US$30 million from funders including former Google CEO Eric Schmidt’s charitable organizations and Skype co-founder Jaan Tallinn. In a blog post Tuesday, Bengio said the group is a response to the rapid advancement in AI and the potential for nefarious behaviours, including deception, cheating, lying, hacking and self-preservation. “I’m deeply concerned by the behaviours that unrestrained agentic AI systems are already beginning to exhibit,” he wrote. Bengio has raised similar concerns in the past, including in the International AI Safety Report published in January that flagged rapid advances like reasoning and agents are adding to the technology’s risks. In response, the Canadian AI Safety Institute said it would heed the report’s warnings through several research projects of its own.