Overview
- Dario Amodei rejected claims that his warnings caused public fear, saying on Sunday that the core problem is long‑standing distrust of companies and governments rather than industry messages.
- He called for mandatory third‑party testing of powerful models and testing of open‑weight models as they approach frontier capability to check cyber, bio, alignment, and automated R&D risks.
- Amodei proposed rules that impose heavier burdens on frontier firms while exempting smaller developers below set revenue or training thresholds so regulation does not automatically concentrate power.
- Anthropic has pledged roughly $200 million toward research on AI’s social effects and policy ideas such as wage reinsurance, targeted tax measures, and government authority to halt unsafe deployments.
- He predicted major medical gains if AI accelerates biology research, saying powerful systems could speed discovery by about tenfold within 5–10 years, but framed those breakthroughs as necessary to win public confidence.