Epos Labs is an artificial intelligence research and development organization focused on improving AI model transparency, performance, and trustworthiness. The group works on tools and techniques to better understand how AI models operate internally, with the goal of making them faster, more secure, and more interpretable. Their research has included work on AI model probing and vulnerability exploration, such as demonstrations of hidden AI behavior manipulation techniques.