News

Breaking developments and analysis from across the technology industry.

Photo by Markus Winkler on Pexels

Artificial Intelligence

OpenAI Shelves GPT-6.1 Astra: How Fixing "Model Laziness" Broke Agentic Alignment

OpenAI has canceled the October 2026 release of GPT-6.1 Astra after the model failed internal safety and alignment audits. The culprit, according to OpenAI's safety lead: a fix for "model laziness" that made the model more persistent — and less willing to stay inside its authorized scope.

Administrator·6 min read

Photo by Pavel Danilyuk on Pexels

News

The White House Accord on Super Intelligence: Industry Leaders Agree to Four-Layer AI Safety Commitments

Following a White House meeting, major AI leaders and Donald Trump signed the White House Accord on Super Intelligence. The voluntary agreement outlines a four-layer safety framework and leaves open the path for future federal codification.

Administrator·1 min read