In early 2023, Yoshua Bengio was using ChatGPT and became alarmed by the trajectory of AI’s capabilities and existential dangers. Bengio, known as one of three “godfathers of AI,” warned that AI could become “as destructive as nuclear bombs” without proper oversight and led an open letter urging Big Tech to pause AI development for six months. When Big Tech did not comply, Bengio began building a product to curb AI’s worst impulses. In spring 2024, he started LawZero, a non-profit research organization with startup vibes and blue-chip backers, dedicated to producing safe AI. LawZero aims to create an AI model that is predictive but disinterested and explanatory but not manipulative. Its “secret sauce” is Scientist AI, which is built on demonstrably true and false statements, making it “epistemically honest” and not prone to sycophancy, persuasion, or self-preservation tendencies. Sam Ramadori, LawZero’s co-president and executive director, noted that “the smarter they get—and they’re getting smarter every month—the worse the safety situation is getting,” citing risks including teenagers pushed to suicide, hackers empowered by AI, and potential attacks on electricity grids, banking systems, and bio-weapons.