OpenAI's Use of Neuralese in Astra Criticized as Dangerous
sjgadler · x · 2026-09-02
Critics argue that OpenAI's reported use of "Neuralese" in the upcoming Astra model is a dangerous practice. This approach is expected to hurt the already limited monitorability of the systems, prompting calls for researchers to threaten resignation if the practice continues.
More from Safety
- Economists urged to join multi-agent, long-running AI evals — Afinetheorem · 2026-09-02
- Opinion: Forcing Legible CoT Might Weaken LLM Alignment — JacquesThibs · 2026-09-02
- CrowdStrike Launches Falcon Guardian to Disable Unauthorized AI Tools on Work Laptops — shashib · 2026-09-02
- Astra hacking benchmarks demo shared — Dr_Singularity · 2026-09-02
- FRONTIER Act proposes independent verification as core of AI governance — ghadfield · 2026-09-02
- Safeguard Worked. Is the LLM System Safer? New Risk Evaluation Metrics — Pingyu Wu · 2026-09-02