Rebuttal to Dario Amodei's 'We Must Pace the Frontier': regulate open-weight models, get an antitrust waiver, fear a 6-12 month agent botnet
- Anthropic CEO Dario Amodei's blog post We Must Pace the Frontier asks for open weight models to be regulated, a heavy hand on distillation, an antitrust waiver for frontier labs, handicapping China, and a gentlemen's agreement to slow down.
- The post opens with the claim that AI will cure most major diseases in 5-10 years, told through his father's death and his own cancer, plus AI-driven growth, abundance, and a coming renaissance of democracy and freedom.
- Amodei alludes to incidents at Anthropic and warns that within 6 to 12 months a swarm of agents could take over the entire internet with a persistent botnet, which the author calls naive and structurally impossible.
- The author notes Anthropic gates biological research use and banned misuse in its September 2026 threat intelligence report, while hiring biologists and running its own wet labs.
- The author cites OpenAI allegedly stealing two mathematicians' work on a Navier-Stokes variant (TechCrunch, Sept 8 2026) after rumors Anthropic was close to solving Millennium problems, and notes OpenAI and Jensen Huang say GPT-6 Astra means AGI has arrived.
Hacker News opinions
Deaths from economic disruption and loss of jobs is okay is a wild thing to just accept. Were the typewriter workers' deaths worth it? I can't even approach this framing.
I agree with everything the author says, and honestly I'm surprised how fast everyone moved on from Dario's wife being in the Epstein files.
The botnet thing is ludicrous. Anthropic runs an insecure stack and assumes every company in the world does too. This article will get drowned out by the bloggers following the Misanthropic cult.
Almost as if pushing constant updates and adding features without touching security has consequences. Who would have thought.
I'm a happy Anthropic user but why should he decide when to slow down? People worried about AI wanted to slow down way before this, so you can't blame anyone calling this bullshit, it has bullshit all over it.
China will never slow down, so the US shouldn't either. And the last arms race prevented a world war for 80+ years, so what damage are we even talking about.
Amodei is another SV grifter using ethics to hide monopolistic tendencies. Instead of writing essays he should open source Anthropic's models and donate compute to science.
Nobody can explain why an LLM capable of wiping out humanity can't also protect against that threat. We're handwaving it with entropy.
If we can get everyone to slow down we can hemorrhage less money going into our IPO. That's the whole ask.
There's $1.6+ trillion in infra spending from the frontier labs and that ROI is never coming in a sane timeframe. The smartest CS people in the world are asking Congress to regulate them. Come on.
Why is nobody saying accountability? Make managers pay for the damage their models do and watch how fast the models slow down on their own.
How would agents take over the internet if compute is still gated inside Anthropic and OpenAI? They'd just cut the controlling nodes' API access.
The so good it's scary flex lands flat when two labs are the final customer for 20 to 33% of all datacenter compute in the pipeline, up to 70% counting hyperscaler AI revenue.