AI labs are becoming more accountable, but not meaningfully more democratic
Saberwing91 · reddit · 2026-07-23
A long Reddit analysis argues that model makers have made some real progress toward making their power more visible and contestable, but not toward sharing meaningful power with the people affected by AI systems.
The author’s ladder of power accounting
The post lays out five levels: rhetoric, legibility, external scrutiny, accountability, and finally power sharing. The claim is that the industry has advanced mostly in the middle rungs, while true power sharing is still missing.
What the author sees as real progress
- The UK AI Security Institute has been given access to frontier models from OpenAI, Anthropic, and Google DeepMind for testing.
- California’s SB 53 adds publishable safety frameworks, critical-incident reporting, whistleblower protections, and civil penalties.
- The EU’s General-Purpose AI Code of Practice and AI Act create stronger oversight, with enforcement powers expected to bite from 2026.
- Anthropic’s Long-Term Benefit Trust can select board members, and OpenAI’s nonprofit structure still formally controls its public-benefit company.
What still looks symbolic or experimental
The post says public-input experiments, such as OpenAI’s democratic-inputs work and Anthropic’s training on preferences from about 1,000 Americans, show that preferences can influence models technically, but do not amount to democratic control.
Bottom line
The author concludes that real institutional changes exist, but they mostly increase visibility and accountability rather than transferring governing or economic power to users, workers, or the public.
More from AGI Musings
- AI safety satire mocks fearmongering over “sandbox escape” claims — max_paperclips · 2026-07-23
- AI may replace junior engineers first, but senior architects still matter — ZeeshanZiaML · 2026-07-23
- More RL may make models look like misaligned agentic systems, author says — zetalyrae · 2026-07-23
- Miles Brundage says AI safety culture has gone too far on blameless postmortems — Miles_Brundage · 2026-07-23
- Miles Brundage says AI safety incidents are broadly predictable even if details aren’t — Miles_Brundage · 2026-07-23
- AI is moving so fast organizations must stay continuously attentive, Peter Slattery says — Exp_Mark · 2026-07-23