Critics accuse Amodei and Altman of the classic playbook: scrape first, discover safety later
TinfoilTricorn · x · 2026-09-15
A widely shared critique argues Amodei and Altman are running Silicon Valley's oldest play: train frontier models on the entire internet's public output — GitHub, Reddit, Wikipedia, blogs — then "suddenly discover ethics and safety" and whisper to Congress about existential risk and the need for regulation.
The core economic argument: both companies are on a treadmill accelerating faster than their revenue curves. Each generation demands exponentially more compute, data and capital, and the previous generation's revenue cannot fund the next training run — so outside capital is required, with the safety narrative serving as bait for investors and policymakers.
It's a pointed but evidence-free opinion piece typical of the AI-safety-criticism genre.
More from AGI Musings
- AI Analyst: 'Join What Becomes' Sentiment Rises as AGI-Only Racing Loses Appeal — danfaggella · 2026-09-15
- tszzl: Backlash against model pacing is an active impediment to the AI safety coalition — tszzl · 2026-09-15
- Penn's new CIS 6280 World Models course publishes all slides, demos and readings — JeremyNguyenPhD · 2026-09-15
- Sentdex: my mom called me to ask about AI x-risk — Sentdex · 2026-09-15
- The question now: which AI leaders haven't picked a side on 'pacing the frontier' — alliekmiller · 2026-09-15
- Ethicist Margaret Mitchell: Don't pause AI, reprioritize from autonomous agents to cancer screening — mmitchell_ai · 2026-09-15