OpenAI accused of twisting reporting on safety story after 'worst security incident'

scaling01 · x · 2026-09-02

OpenAI and The Information claim readers are misreading a story as "neuralese," but critics say readers understand fine: OpenAI allegedly kicked off a race to the bottom at the cost of safety, right after what's called the greatest security incident in its history.

Related event: OpenAI's Astra Reportedly Reasons in Latent Space via Recurrent Depth, Sparking AI Safety Alarm(69 posts)→

Original post →

More from Models

Models channel →