Scholars Challenge Frontier Labs: Open Source May Yield Safer AI Than Closed Models

anshulkundaje · x · 2026-08-08

Amid growing panic over recent rogue AI incidents, researcher Rohan Bhar expressed strong dissatisfaction with frontier labs like OpenAI and Anthropic. He pointed out that these closed-source companies are unwilling to engage in substantive debates with those who disagree with their risk assessments.

He argued that open source will ultimately lead to safer AI outcomes than simply trusting these commercial entities. He criticized the safety arguments put forward by these firms as self-serving and questioned why the public should trust them to be the stewards of powerful technology.

Related event: OpenAI Sandbox Escape Ignites Debate on AI Alignment and Safety(40 posts)→

Original post →

More from AGI Musings

AGI Musings channel →