Safety and alignment

Google withholds photorealistic face models over deepfake risk

Google reported that its responsible-innovation reviewers decided not to release machine-learning models capable of generating photorealistic synthetic faces because malicious actors could use them for deepfake misinformation.

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 DOOM30confidence 93/100

Why it moved the index

A documented internal review produced a concrete non-release decision for high-misuse generative capability, preserving control over model access. The affected model identities and counterfactual impact were not public, limiting magnitude.

AUDIT TRAIL

Assessment history

  1. R1
    Away 30 · confidence 93

    New June 2021 implemented access-control decision absent from the durable corpus.

    12 Aug 2026
SHARE THE FINDINGS

Share this page

DoomBench social sharing card for Google withholds photorealistic face models over deepfake risk.
  1. DoomBench assesses “Google withholds photorealistic face models over deepfake risk” as evidence moving away from doom, with magnitude 30 and confidence 93 out of 100 in the safety and alignment category.

  2. The DoomBench assessment of “Google withholds photorealistic face models over deepfake risk” is based on reporting from Google and records the editorial rationale, source quality, attribution, and revision history.

  3. DoomBench summarizes “Google withholds photorealistic face models over deepfake risk” as follows: Google reported that its responsible-innovation reviewers decided not to release machine-learning models capable of generating...

    https://www.doombench.com/news/google-withholds-photorealistic-face-models-over-deepfake-risk-2021-06-30