Thomas Woodside on Require AI labs to publish safety evalu…

Thomas Woodside

Secure AI Project co-founder

Those updates should include the results of evaluations for models that haven’t been publicly deployed yet, since those models could also pose serious risks.

AI Verified source (2025)

7mo ago

Policy proposals and claims

Require AI labs to publish safety evaluations before deploying frontier models

votes For

Statement relation verification history AI Verified Report this

AI Verified The quote supports publishing evaluation results before deployment: it says updates "should include the results of evaluations for models that haven’t been publicly deployed yet." That clearly implies support for pre-deployment publication of safety evaluations, especially for risky/frontier models. · YouCongress gpt-5.4-2026-03-05 · 3mo ago

Vote inference verification history AI Unverifiable Report this

AI Unverifiable The quote says updates "should include the results of evaluations for models that haven’t been publicly deployed yet," which suggests publishing some pre-deployment eval results, but it does not clearly address a formal requirement on AI labs or specifically frontier-model safety evaluations. · YouCongress gpt-5.4-2026-03-05 · 3mo ago

Quote authenticity verification history

Report this

AI Verified ai_verified: The TechCrunch article dated April 17, 2025 contains the exact sentence provided at line 131 and explicitly attributes it to Thomas Woodside as a statement he made to TechCrunch. ([techcrunch.com](https://techcrunch.com/2025/04/17/googles-latest-ai-model-report-lacks-key-safety-details-experts-say/)) · YouCongress gpt-5.4-2026-03-05 · 3mo ago

replying to Thomas Woodside

Other opinions from Thomas Woodside

See all

Thomas Woodside votes For

AI companies should be liable for harms caused by their deployed models

Liability already exists under common law and provides a powerful incentive for AI companies to take reasonable steps to prevent foreseeable risks from their AI systems. SB 3444 would take the extreme step of nearly eliminating liability for severe h...

more AI Verified source (2026)