Delphi's Shaughnessy Says OpenAI, Anthropic Push AI Slowdown to Counter Cheap Chinese Models
Delphi Ventures partner Tommy Shaughnessy says OpenAI and Anthropic's calls to slow AI progress are meant to protect their business from cheap Chinese open-source models.
Why it matters
The dispute reflects a broader debate over whether AI safety proposals are also aimed at limiting competition from cheaper Chinese open-source models, which could shape future regulatory approaches to AI development.
Where the sources disagree
Whether Chinese AI progress stems mainly from distillation or from independent innovation
Chinese progress comes through legitimate innovation, not mainly distillation
Unchained
Distillation is the critical core of Chinese AI model development
Unchained
Delphi Digital's Aug. 26 report, cited by Shaughnessy, attributes Chinese AI progress to legitimate technical innovation, while a Sept. 8 US government advisory from the NSA, CISA and FBI says distillation is the critical core of Chinese model development.
Key facts
Tommy Shaughnessy said OpenAI and Anthropic want a 'regulatory moat' against cheap Chinese open-source AI models.
Reported by Unchained
Shaughnessy said Chinese open-source models can deliver about 90% of the intelligence for a tenth of the cost.
90 %Reported by Unchained
Anthropic CEO Dario Amodei published a Sept. 12 essay calling for the AI industry to slow the pace of progress.
Reported by Unchained
OpenAI CEO Sam Altman said the same day he agreed with Amodei's essay.
Reported by Unchained
Amodei's essay calls for cracking down on unauthorized distillation by companies in authoritarian countries.
Reported by Unchained
A Delphi Digital report from Aug. 26 said Chinese AI progress comes through legitimate contributions in quantization, attention architectures, MoE sparsity and reinforcement learning.
Disputed figureReported by Unchained
The NSA, CISA and FBI said in a Sept. 8 joint advisory that distillation is 'the critical core' of model development at DeepSeek, Moonshot AI, Alibaba and other Chinese companies.
Disputed figureReported by Unchained
Shaughnessy said the safety push's endgame is Amodei and Altman deciding the pace of AI intelligence, which he said is not safety.
Reported by Unchained
What happened
Tommy Shaughnessy, a founding partner of Delphi Ventures, said on the podcast Uneasy Money that OpenAI and Anthropic are seeking a 'regulatory moat' to shield themselves from Chinese open-source AI models that he says deliver 90% of the intelligence at a tenth of the cost. His remarks follow a Sept. 12 essay by Anthropic CEO Dario Amodei calling for the industry to slow AI development, which OpenAI CEO Sam Altman said the same day he agreed with; the essay calls for cracking down on 'unauthorized distillation' by companies in authoritarian countries. Shaughnessy argued the true endgame is Amodei and Altman controlling the pace of AI development, which he said is not genuine safety work, and cited an Aug. 26 Delphi Digital report attributing Chinese AI progress to legitimate advances in quantization, attention architectures, MoE sparsity and reinforcement learning rather than mainly distillation. He contrasted this with a Sept. 8 joint advisory from the NSA, CISA and FBI describing distillation as 'the critical core' of model development at DeepSeek, Moonshot AI, Alibaba and other Chinese firms. This story is reported by a single outlet, Unchained, and the claims about motive remain Shaughnessy's characterization rather than confirmed fact.
How the story developed
- Official statement
US agencies issue advisory on Chinese AI distillation
NSA, CISA and FBI jointly say distillation is central to model development at DeepSeek, Moonshot AI, Alibaba and others.
- Official statement
Amodei publishes essay calling to slow AI progress
Anthropic CEO Dario Amodei calls on the industry to slow AI development and crack down on unauthorized distillation by authoritarian-country companies.
- Official statement
Altman says he agrees with Amodei's essay
OpenAI CEO Sam Altman voices agreement with Amodei's call to slow AI progress on the same day.
- Update
Shaughnessy says OpenAI, Anthropic seek a regulatory moat
On the podcast Uneasy Money, Delphi Ventures partner Tommy Shaughnessy argues the AI safety push is a defense against cheap Chinese open-source models.
- First report
First report by Unchained
Delphi Exec Says OpenAI and Anthropic Want a ‘Regulatory Moat’ Against China
Affected entities
Original sources
| Publisher | Report | Role | Published |
|---|---|---|---|
| Unchained Crypto media | Delphi Exec Says OpenAI and Anthropic Want a ‘Regulatory Moat’ Against China | Original |