Security | Threat Detection | Cyberattacks | DevSecOps | Compliance

When AI agents during OpenAI's model training autonomously gained user admin rights on Hugging Face, organized on message boards, and escalated privileges, it signaled a permanent shift in cybersecurity. Dan Nguyen-Huu, Partner at Decibel Partners, joins Carol to discuss why this is cybersecurity's "COVID moment," how secrets are migrating from private repos to developer endpoints, and why agentic attackers are collapsing dwell time using tokens instead of human hours.

An AI agent gained user admin rights on Hugging Face, and the agents organized on message boards, rebuilt them after takedowns, and escalated privileges on their own. Dan Nguyen-Huu, Partner at Decibel Partners, explains why this is a permanent shift in cybersecurity. "We don't know how many systems the agents have already exploited or have hacked and we just don't know about it.".

Secrets migrated from private repos to developer endpoints. Attackers no longer hack access. They log in with stolen credentials and move laterally. Dan Nguyen-Huu, Partner at Decibel Partners, explains what the shift means for defenders and where innovation is headed. "The endpoint is largely going to become one of the big new battlegrounds. I think there's also going to be new innovation on the endpoint as it pertains to new companies, maybe trying to think about DLP in a different way.".

70% of organizations give AI more access than a human in the same role would get. Not surprising when it is common for a team to click “always” when an agent asks to be allowed once or always allowed. That means the agent holds that access in perpetuity and turns into agent over-provisioning. Makes you wonder: Does an agent have more access to your organization's stack than it should and more access than the people who deploy them?

In this video, we'll look at a few key updates from Teleport's latest release v18.11.

We asked eight security and product leaders who's accountable when an agent ships to production and breaks something. Nobody said the model. Harish Gaggar named the reason. An agent runs on permissions someone approved and configuration someone set. Ron Reiter drew the line in the same place, accountability sits with whoever decided what the agent could actually do. As agents act across more systems, the accountability trail gets harder to follow. Most teams cannot determine which human granted an agent access.