Y Combinator CEO Garry Tan has publicly advocated for allowing American open-weight AI labs to distill proprietary frontier AI models, pushing back against regulatory interventions aimed at restricting the practice. Speaking to CNBC and TechCrunch, Tan urged US regulators to maintain open policies, arguing that establishing an domestic distillation framework would ensure American open-weight models remain competitive against foreign developments.
Tan’s comments follow a report from Anthropic alleging that Chinese labs are conducting unauthorized distillation attacks using fraudulent credentials. Anthropic CEO Dario Amodei previously urged US officials to restrict unauthorized distillation. While Tan clarified that he does not support using stolen credentials, he argued that closed-model providers should not overly restrict how customers utilize API outputs, noting that proprietary models were themselves trained on broadly accessible public data.
Emphasizing the importance of ecosystem balance, Tan warned that limiting open-weight development risks creating a monopolistic industry dominated by a single well-capitalized frontier lab. He argued that public access to intelligence should be treated as a public good rather than locked behind restrictive corporate terms of service.
Why it matters
Venture leaders are actively opposing API usage restrictions that limit how open-weight startups fine-tune on closed models.
Regulatory decisions on distillation will directly shape the competitive dynamic between proprietary frontier labs and open-source startups.
Source: techcrunch.com
