Safety and alignment

OpenAI dedicates 20% of secured compute to superalignment

OpenAI formed a four-year Superalignment team and committed 20% of its secured compute to controlling systems smarter than humans, explicitly citing disempowerment and extinction risks.

0 comments · 0 votesOpen discussion

Public discussion is readable by everyone. Sign in to comment, reply, or vote.

No comments yet. Start the discussion.

CURRENT ASSESSMENT · REVISION 1
AWAY FROM DOOM58confidence 96/100

Why it moved the index

A frontier laboratory committed substantial compute and a dedicated team to the direct technical problem of retaining human control over superintelligent systems, although success remained uncertain and future-facing.

AUDIT TRAIL

Assessment history

  1. R1
    Away 58 · confidence 96

    New July 2023 direct superintelligence-control investment milestone.

    12 Aug 2026
SHARE THE FINDINGS

Share this page

DoomBench social sharing card for OpenAI dedicates 20% of secured compute to superalignment.
  1. DoomBench assesses “OpenAI dedicates 20% of secured compute to superalignment” as evidence moving away from doom, with magnitude 58 and confidence 96 out of 100 in the safety and alignment category.

  2. The DoomBench assessment of “OpenAI dedicates 20% of secured compute to superalignment” is based on reporting from OpenAI and records the editorial rationale, source quality, attribution, and revision history.

  3. DoomBench summarizes “OpenAI dedicates 20% of secured compute to superalignment” as follows: OpenAI formed a four-year Superalignment team and committed 20% of its secured compute to controlling systems smarter than humans, explicitly...

    https://www.doombench.com/news/openai-dedicates-20-of-secured-compute-to-superalignment-2023-07-05