← Retour au blog
tech 13 September 2026

Garry Tan Wants US AI Labs to Also 'Distill' Frontier Models

Garry Tan, CEO of Y Combinator, suggests that US AI labs adopt distillation to bolster their competitiveness against China. But what does this really entail?

Article inspired by the original source
Garry Tan wants US open-weight AI labs to 'distill' frontier models, too ↗ techcrunch.com

Why is Distillation Crucial?

Model distillation is a key technique in AI development. It enables transferring knowledge from a complex model to a simpler one, maintaining similar performance. This method offers several benefits: cost reduction, improved execution speed, and increased accessibility to advanced technologies.

Garry Tan, CEO of Y Combinator, highlights the importance of this technique for US AI labs. According to him, adopting distillation could provide the US with a robust alternative, distinct from the growing Chinese influence in this field. It's not about encouraging illicit practices, but legitimizing and fostering innovation through transparent and open means.

Concerns About Current Practices

Recent accusations by Anthropic, a specialized AI company, highlight that some Chinese labs may use distillation techniques without permission. These "illicit distillation attacks" reportedly rely on hidden identities and stolen information. While this might seem alarming, Tan argues that US labs should be allowed to explore distillation in legitimate and ethical ways.

It's important to note that many AI labs themselves have used data without permission to train their models, including copyrighted content. Tan argues it would be hypocritical to restrict what customers can do with the information shared by these models.

The Opportunity for US Labs

By encouraging distillation, Tan envisions a new wave of innovation in the US. This could allow smaller labs to compete with giants like OpenAI and Google. Distillation could also facilitate the creation of more specialized models better suited to specific needs.

According to a 2023 McKinsey report, the AI market is expected to reach $390 billion by 2025. By leveraging distillation, US companies could capture a larger share of this booming market.

How to Implement Ethical Distillation?

For distillation to be adopted ethically, it is crucial to establish clear standards and guidelines. This includes obtaining appropriate consents and ensuring complete transparency in model training processes.

Companies should also invest in R&D to improve distillation techniques, ensuring they don't compromise quality or ethics. Ultimately, it's about creating an ecosystem where innovation and ethics coexist harmoniously.

Conclusion

Model distillation represents a strategic opportunity for US AI labs. By adopting this technique, they can not only strengthen their competitiveness but also drive innovation ethically. Tan reminds us that the future of AI requires both boldness and responsibility.

Let's discuss your project in 15 minutes.

AI distillation Garry Tan US AI labs ethical AI innovation
Deepthix newsletter · 100% AI · every Monday 8am

An AI agent reads tech for you.

Our AI agent scans ~200 sources per week and ships the best articles to your inbox Monday 8am. Free. One click to unsubscribe.

Visit the newsletter page →

Want to automate your operations?

Let's talk about your project in 15 minutes.

Book a call