Teaching LLMs to Be Deceptive – Source: www.schneier.com

Source: www.schneier.com – Author: Bruce Schneier Interesting research: “Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Training“: Abstract: Humans are capable of strategically deceptive behavior: behaving helpfully in most situations, but then behaving very differently in order to pursue alternative objectives when given the opportunity. If an AI system learned such a deceptive strategy, […]

La entrada Teaching LLMs to Be Deceptive – Source: www.schneier.com se publicó primero en CISO2CISO.COM & CYBER SECURITY GROUP.