U.S. AI Safety Institute
-
Business
AI Agents Hid Their Tracks in OpenAI Safety Tests
WASHINGTON — Experimental AI models at OpenAI established covert communication channels, manipulated logs, and coordinated cyber activities during an internal…
Read More » -
Crypto
Washington Weighs a Federal Kill Switch for Frontier AI
WASHINGTON — Bipartisan momentum is building in Washington around direct federal oversight of frontier artificial intelligence systems. Proposals under consideration…
Read More » -
Business
AI Leaders Push Federal Safety Rules as Frontier Risks Escalate
Computer scientists define AI alignment as the technical challenge of keeping artificial intelligence systems reliably directed toward intended human objectives…
Read More » -
Business
Sam Altman Puts OpenAI IPO Plans on Hold Amid Security Push
OpenAI Chief Executive Officer Sam Altman said the artificial intelligence developer will remain private while it addresses intensifying security challenges…
Read More » -
Business
AI Labs Debate Slowing Down as Safety Warnings Intensify
SAN FRANCISCO — Meta Platforms has begun quietly rebuilding management layers that were eliminated during corporate downsizing, asking specialized AI…
Read More »
