In a recent social media post, Anthropic CEO Dario Amodei highlighted a "crisis of trust" surrounding AI, stating that the public's negative perception is a significant problem for the industry. [6] His comments double down on his consistent call for robust government oversight of powerful AI systems, which he believes is necessary to ensure public safety as the technology advances rapidly. [1, 16]
Amodei's position distinguishes between different approaches to AI development. While he does not support a complete ban on open-weights models, he argues they are not a "sufficient" safety solution and that broad access does not necessarily help defenders more than attackers. [5, 7] He advocates for mandatory, FAA-style safety testing and auditing for all highly capable models, both open and closed, to manage potential catastrophic risks. [1, 17]