Watchdog nonprofit tracking whether AI companies keep their safety promises.
The Midas Project is a small watchdog that keeps score on AI companies' safety pledges and publishes when they slip. Its trackers are the main public record of the Seoul commitments' follow-through.
Focus
- Transparency
- Safety evaluations
- Anti concentration
Funding, as far as we know
Donations (501(c)(3), EIN 99-2326977); specific funders unknown - https://www.themidasproject.com/projects
What they've actually done
Launched the Seoul Commitment Tracker grading 16 companies; found only 6 fully met the Feb 2025 deadline
28 February 2025
Documented xAI missing two self-imposed deadlines for a safety framework
13 May 2025
Published 'The OpenAI Files' with the Tech Oversight Project, cataloguing broken promises and governance concerns
18 June 2025
Was subpoenaed by OpenAI in the Musk litigation; runs the 'Watchtower' log of safety-policy changes (latest entries Aug 2026)
2025-09 / 2026-08