xAI's Risk Framework Mentions Internal Capability Surveys, But Outcomes Remain Unclear

dfrsrchtwts · x · 2026-07-23

A discussion highlights that xAI's risk management framework mentions potentially publishing survey results from employees regarding their views and projections on important future AI developments, such as capability gains and benchmark results. The poster inquires whether this has actually happened, noting that the closest comparable industry practice so far is Anthropic publishing surveys on how their models perform on AI R&D tasks.

Related event: xAI Risk Framework Mentions Internal AI Prediction Survey, Results Remain Unreleased(2 posts)→

Original post →

More from Safety

Safety channel →