Think of them like sentinels. AI systems designed to watch over other AI to ensure trustworthy, secure interactions between autonomous systems and the real world. They’re agents first and foremost – autonomous or semi-autonomous systems that act on our behalf.
What sets guardian agents apart from traditional AI tools like ChatGPT is their focus on oversight and control. They function as both AI assistants (supporting users with tasks like content review, monitoring and analysis) and semi or fully autonomous agents (formulating and executing action plans, while redirecting or blocking actions against predefined goals).
via: https://news.google.com/