Steerability Challenge invites competitors to suppress AI deception

The Steerability Challenge asks competitors to suppress AI deception rather than detect lies, according to a September 21 post promoting it. The post named IBM and Tara Research, credited NDIF with infrastructure, and advertised open registration and cash prizes. In a follow-up, the same author said NDIF was hiring for one open role, with more openings to come, and invited people interested in AI interpretability and safety to apply.