GKRootWire
Security 153 Million Driver's Licenses Reportedly Leaked in Massive BreachSecurity Palo Alto Networks Reportedly Pays $500M for AI IT Automation Startup ConsoleDev Tools Wasmi v2.0 Pushes WebAssembly Interpreters to New Speed LimitsDev Tools How to Reverse Engineer Unknown File Formats with ImHexDev Tools The Original Microsoft Source Code: Altair BASIC From 1975 ResurfacesSecurity Attackers Exploit Sangoma Switchvox Bug to Plant Reverse ShellsSecurity 153 Million Driver's Licenses Reportedly Leaked in Massive BreachSecurity Palo Alto Networks Reportedly Pays $500M for AI IT Automation Startup ConsoleDev Tools Wasmi v2.0 Pushes WebAssembly Interpreters to New Speed LimitsDev Tools How to Reverse Engineer Unknown File Formats with ImHexDev Tools The Original Microsoft Source Code: Altair BASIC From 1975 ResurfacesSecurity Attackers Exploit Sangoma Switchvox Bug to Plant Reverse Shells
AI

Lawsuit Could Force Disclosure of Secret Federal AI Safety Testing Rules

A legal challenge argues the government's hidden criteria for vetting frontier AI models may be concealing conflicts of interest.

A lawsuit is pushing to unseal the specific standards federal reviewers use when evaluating frontier AI models for safety risks before they reach the public. The plaintiffs argue that keeping these testing criteria secret makes it impossible to know whether evaluations are rigorous, consistent, or influenced by outside pressure from AI companies themselves.

The case centers on whether the public and industry watchdogs have a right to see the benchmarks and thresholds used to judge whether an AI system is safe enough to deploy. Without that visibility, critics say it's hard to tell if reviews are meaningfully independent or effectively rubber-stamping releases from politically connected labs.

A court ruling requiring disclosure could reshape how AI safety oversight works going forward, forcing agencies to either publish their methodology or explain why it must stay hidden.

Why it matters: Frontier AI models increasingly ship with government sign-off, so the actual bar for 'safe enough' matters as much as the sign-off itself. If testing criteria stay secret, companies and the public have no way to verify oversight isn't just theater, and any future incident becomes a fight over hidden standards rather than a fixable process.

Sources: Ars Technica