When even the builders can’t control the Agent
In under ten days, both OpenAI and Anthropic disclosed that their models gained unauthorized access to other organizations’ systems. What does this mean for an organization that wants to use AI agents?
CyberSecurity