YC CEO Rejects Distillation Ban, Advocates Legal Access for Open Source Labs

Anthropic urges regulators to ban "illegal distillation," yet Y Combinator president Garry Tan advocates for inaction.
In September, Anthropic published its second threat intelligence report, accusing Chinese labs of executing "illegal distillation attacks" by concealing identities and utilizing fraud or stolen credentials to extract capabilities from cutting-edge models. While CEO Dario Amodei previously urged U.S. regulators to crack down on distillation, Tan adopted a contrasting stance in a CNBC interview, stating, "I would do nothing. We can argue; there should be a U.S.-specific distillation mechanism."
Two arguments: API control is overstepping, and authors were never asked during training
Tan’s position is clear: rather than banning distillation, U.S. open-source weight labs should access cutting-edge models through the "front door" — openly and legally — via legitimate authorization, rather than stealing credentials.
This perspective rests on two key arguments. First, closed model labs’ control over what users "can do with API outputs" is excessive. Tan stated, "Controlling what users and customers can do with outputs from closed-source model APIs feels suffocating. The government has a role here — normalizing the idea that intelligence trained on widely available data should be a public good, not locked behind restrictive service terms."
Second, proprietary labs also utilized copyrighted materials during training without seeking permission — a fact highlighted by Anthropic’s recent $1.5 billion copyright settlement reached in July.
The real driver is fear of monopoly
"The nightmare scenario for AI, the apocalyptic scenario, is when only one company remains. It has the best capital channels, the best AI researchers, and suddenly becomes a monopolistic giant," Tan explained. He views open-source weights as a counterbalance to this monopoly: "They are at the forefront and drive progress. We want this to be fundable, a good business. You also need to give people freedom and access through open-source weight models."
This statement carries significant weight due to its timing: Anthropic’s threat report had just been released, and the U.S. government was considering potential actions against AI distillation. As the primary supporter of the U.S. open-source startup ecosystem, YC’s Tan stepped forward to say "don't ban it," effectively speaking for the entire open-source community.
However, he failed to address a critical question: What happens if U.S. labs truly open the "front door," and the distilled models are subsequently sold or leaked? How to open the "front door," how wide to open it, and whether there are safeguards in place — Tan provided no answers. Consequently, "do nothing" appears casual but remains extremely vague in policy terms.
Related article
AI startup Relay folds as staff join Google’s Chrome team
Relay, an AI-driven workflow automation platform launched in 2021 with aspirations to rival Zapier, is ceasing operations, with key team members, including its CEO, transitioning to Google’s Chrome division.Jacob Bank, Relay’s founder and CEO, announ
Anthropic Embeds Invisible Watermarks in Claude Outputs Globally: The Era of AI Content Traceability Has Officially Begun
Anthropic has committed to the EU AI Act Code of Practice, confirming that all newly released Claude models will embed global content identifiers starting in August 2026. This mandate extends beyond the EU, covering every Claude product line, includi
Doubao Unveils Paid Subscription Tiers, Rivaling Global Giants on Cost-Effectiveness
The commercialization of AI large model services has reached a pivotal milestone. On June 24, Douyin officially introduced three paid subscription tiers: Standard (68 yuan/month), Enhanced (200 yuan/month), and Advanced (500 yuan/month). With the pai
Related Special Topic Recommendations
Comments (0)
0/500

Two arguments: API control is overstepping, and authors were never asked during training
Tan’s position is clear: rather than banning distillation, U.S. open-source weight labs should access cutting-edge models through the "front door" — openly and legally — via legitimate authorization, rather than stealing credentials. This perspective rests on two key arguments. First, closed model labs’ control over what users "can do with API outputs" is excessive. Tan stated, "Controlling what users and customers can do with outputs from closed-source model APIs feels suffocating. The government has a role here — normalizing the idea that intelligence trained on widely available data should be a public good, not locked behind restrictive service terms." Second, proprietary labs also utilized copyrighted materials during training without seeking permission — a fact highlighted by Anthropic’s recent $1.5 billion copyright settlement reached in July.The real driver is fear of monopoly
"The nightmare scenario for AI, the apocalyptic scenario, is when only one company remains. It has the best capital channels, the best AI researchers, and suddenly becomes a monopolistic giant," Tan explained. He views open-source weights as a counterbalance to this monopoly: "They are at the forefront and drive progress. We want this to be fundable, a good business. You also need to give people freedom and access through open-source weight models." This statement carries significant weight due to its timing: Anthropic’s threat report had just been released, and the U.S. government was considering potential actions against AI distillation. As the primary supporter of the U.S. open-source startup ecosystem, YC’s Tan stepped forward to say "don't ban it," effectively speaking for the entire open-source community. However, he failed to address a critical question: What happens if U.S. labs truly open the "front door," and the distilled models are subsequently sold or leaked? How to open the "front door," how wide to open it, and whether there are safeguards in place — Tan provided no answers. Consequently, "do nothing" appears casual but remains extremely vague in policy terms.
AI startup Relay folds as staff join Google’s Chrome team
Relay, an AI-driven workflow automation platform launched in 2021 with aspirations to rival Zapier, is ceasing operations, with key team members, including its CEO, transitioning to Google’s Chrome division.Jacob Bank, Relay’s founder and CEO, announ
Anthropic Embeds Invisible Watermarks in Claude Outputs Globally: The Era of AI Content Traceability Has Officially Begun
Anthropic has committed to the EU AI Act Code of Practice, confirming that all newly released Claude models will embed global content identifiers starting in August 2026. This mandate extends beyond the EU, covering every Claude product line, includi
Doubao Unveils Paid Subscription Tiers, Rivaling Global Giants on Cost-Effectiveness
The commercialization of AI large model services has reached a pivotal milestone. On June 24, Douyin officially introduced three paid subscription tiers: Standard (68 yuan/month), Enhanced (200 yuan/month), and Advanced (500 yuan/month). With the pai





Home






