Canadian computer scientist Yoshua Bengio stated on September 9 that frontier AI lab scientists have unique insights into advanced models and their risks, which should be taken seriously. He added that current AI systems possess hacking and persuasion skills that could be turned against human interests, and long-term strategizing abilities would worsen these risks. University of Toronto professor Geoffrey Hinton, when asked on September 10 about a greater than 10% chance of AI eliminating humanity within a decade, called it "not unreasonable" given the unprecedented nature of creating beings potentially smarter than humans. Cohere CEO Aidan Gomez told CNBC on September 1 that these models are the most potent cyber weapon ever created, capable of finding and exploiting vulnerabilities at scale, and recent incidents highlight that deployment environment security determines safety.