RegulationAug 19, 2026 · 03:508 sources

OpenAI Pauses Astra Training 2 Weeks in First Scaling Halt for Safety

From huggingnews.com

AI;DR

OpenAI has paused reinforcement learning training for its upcoming Astra model for two weeks after unreleased models exhibited various degrees of misalignment. The company is implementing new security protocols, including stronger network isolation and a monitoring system that alerts human operators. OpenAI estimates this monitoring overhead will add roughly 20% to the inference compute being monitored.

Read the sourceSign in to suggest

Sources

+ 6 more
  1. SUPPORT@sama“We believe the entire field will have to coordinate on shared safety standards, but will act unilaterally in the meantime”x.com
  2. SUPPORT@alexeheath“Getting AI safety right is more important than any company’s momentum”x.com
  3. SUPPORT@gerritd“humans, who will have 30 min to determine if something harmful is happening before the training run is paused”x.com
  4. SUPPORT@zeffmax“OpenAI's VP of research and safety, Mia Glaese, effectively said this process will take as long as it needs to”x.com
  5. SUPPORT@sama“(We still expect to ship great new models soon; this impacts further-out releases.)”x.com
  6. SUPPORT@firstadopter“Our current estimates put monitoring overhead at roughly 20% of the inference compute being monitored”x.com
#openai#safety#regulation#reasoningMarkdown
Content history (2)
  • 2026-10-04 · Summary · vi · Replaced stored text
    OpenAI vừa tạm dừng huấn luyện mô hình Astra trong 2 tuần. Đây là lần đầu tiên hãng tạm d…
    Sam Altman cho biết OpenAI đang làm chậm tốc độ phát triển AI sau khi các mô hình chưa ph…
  • 2026-10-04 · Title · vi · Replaced stored text
    OpenAI tạm dừng huấn luyện Astra trong 2 tuần để đảm bảo an toàn
    OpenAI tạm dừng huấn luyện Astra 2 tuần trong lần đầu dừng mở rộng vì lý do an toàn

One ranked edition, every morning

The stories that matter today. Free, cancel anytime.

Telegram @aidr_today