I can only sigh and shake my head. If we had a capable, responsible, well structured government... then Amodei's essay would make sense.
We don't. We have a President who is a former reality TV star, who surrounds himself with dangerous sycophants. We have a congress frequently guilty of insider trading and unaccountable. We have unlimited, legalized system of political bribery thanks to the Citizens United ruling of 2010.
The problem is that this government is corrupt and broken at every level.
They cannot be expected to manage AI regulation. They cannot manage a reasonable budget, foreign policy, or social support system. Our infrastructure, broadly, is failing at a level that looks likely to become Catastrophic.
Amodei's proposals might otherwise make sense. Might otherwise be reasonable.
But Anthropic itself is part of the problem. The mass profit motive, the weaponzation and - yes - surveillance their systems are already used for. The objection camd rather too late.
And now, this. A framework for control, for regulation. Not a terrible one, but not a good one either.
Let me be blunt: it is dead on arrival.
The oligarchs, the lobbyists, some of them even affiliated with anthropic... will decide whether any real regulatory framework appears. Not the federal government. That is a broken institution with no more checks and balances. That can no longer serve its intended purpose. That hasn't, for some time.
Until or unless that changes, I will personally oppose AI regulation. No one I believe is trustworthy is working on it. Frankly, I think this is good PR, but hopeless in terms of actually going anywhere.
Anthropic is already degrading ML and AI model research quietly, through Claude Fable. That was made clear in the system prompt many of us saw yesterday.
So, no. I don't trust Anthropic. I think Amodei has good intentions, but his motivations make him too suspect to lead on this, either.
Maybe you'll get some support.
But I'll say again: DOA. That's simply the truth.