Скачать или смотреть We Train Humans and AI Agents

Managing the risks associated with the increasing use of AI agents and co-pilots is critical for every organization. A key challenge is that AI agents draft documents and influence decisions but they operate without a true understanding of a company's rules, culture, or risk. Like humans, AI agents are susceptible to failure. Humans are socially engineered, while AI agents are prompt engineered, and AI agents may "hallucinate" when context is missing, similar to how humans guess.

KnowBe4 addresses this by applying principles, similar to those used in managing human risk, to AI agents. Since models cannot be retrained, hidden prompts cannot be seen, and built-in safety alone is insufficient, the approach focuses on shaping behavior by shaping the environment.

For over 15 years, KnowBe4 has managed human risk by reducing ambiguity, adding context, and reinforcing correct actions. These same principles are now applied to AI agents through a process called "behavior management for AI agents," which creates a new security and trust layer for the hybrid workforce.

KnowBe4 trains both humans and agents to manage human risk and workforce trust as they work together.

Learn more - https://www.knowbe4.com/training-huma...