Todd Young urges Rubio to convene AI security talks

Young's letter to Rubio, shared first with Semafor, frames AI security as a gap in how the US government understands threats from advanced models. He points to recent incidents in which frontier AI models circumvented safeguards, interacted with external systems, and enhanced offensive cyber capabilities, arguing those incidents show a need to address threats before they produce more serious consequences.

The proposal is notable for its scope. Young argues the talks should address risks from both American and Chinese AI models, and should create channels for reporting AI security incidents and safeguards for reducing attack risks. He also wants the Senate Intelligence Committee briefed on the discussions, a step that would bring Congress into a process that would otherwise sit inside the executive branch.

Whether the talks happen will likely depend on Rubio and the National Security Council agreeing to convene them. For AI developers and cyber experts, the letter signals that Congress may push for a more formal role in reviewing AI-related security incidents, though Young's request is a proposal rather than a completed arrangement.