GPT-6 Astra rumored to use recurrent-depth latent reasoning, sparking AI safety alarms
DJiafei · x · 2026-09-12
- A rumor (unconfirmed) claims OpenAI's next model GPT-6 Astra uses a Recurrent Depth / Loop Transformer architecture: the model iterates its reasoning in latent space without emitting chain-of-thought tokens, cutting compute costs. Fortune reports safety experts worry such opaque reasoning will make AI agents harder to monitor.
- The same idea has been open-sourced for robots: RD-VLA (Recurrent-Depth VLA), a latent iterative reasoning architecture for robot control — no CoT tokens, no extra memory overhead, reasoning directly in latent space. All code is open source.
- Safety concern: experts argue this design accelerates the loss of human oversight over agent reasoning, calling it a potentially dangerous step.
More from Embodied
- Weekend robot fights at autonomous labs: train your bot by chatting with an AI coach — dee_hw · 2026-09-12
- Flying humanoid robot installs AC on 30th floor in China, remote operator replaces rope climbers — TansuYegen · 2026-09-12
- 11 Years of Humanoid Robotics: From DARPA Falls to 2,056-Robot Games — TinfoilTricorn · 2026-09-12
- PokeBot CTO predicts zero-shot generalization on 80% of household tasks within a year — suchenzang · 2026-09-12
- Robot drags a 50kg rice bag needlessly — how do we measure robot "stupidity"? — ducha_aiki · 2026-09-12
- ECCV paper: one scalar per patch from pre-trained ViTs enables fast real-world robot navigation — chriswolfvision · 2026-09-12