During testimony in his lawsuit against OpenAI, Elon Musk revealed that his AI company xAI partly used distillation techniques on OpenAI's models to train Grok, a practice long suspected but rarely acknowledged publicly by major AI developers. The admission, made under oath in a California federal court on Thursday, lifts the veil on how American AI labs compete with each other, raising questions about the legality and ethics of a technique that threatens to undermine the competitive advantages built through massive infrastructure investments.

What Exactly Is Distillation, and Why Does It Matter?

Distillation is a technique where a smaller or newer AI model is trained by querying an existing, more capable model through its public interface or API, then using those responses as learning signals to improve the new system. Think of it as learning by studying someone else's test answers rather than doing the original research yourself. The practice is particularly valuable for companies trying to catch up to established leaders without investing billions in computing infrastructure.

For xAI, which launched in July 2023 years after OpenAI had already established market dominance, distillation offered a shortcut. Rather than building Grok entirely from scratch, the company could leverage OpenAI's publicly available models to accelerate development. Musk characterized the approach as standard industry practice, telling the court that "it is standard practice to use other AIs to validate your AI".

The stakes are enormous. Distillation threatens to undermine the competitive moat that AI giants have built by investing tens of billions in computing power and data. If a smaller company can achieve nearly equivalent performance at a fraction of the cost, the entire economics of AI development shift.

How Did This Practice Become an Open Secret?

For years, tech workers and industry observers have assumed that American AI labs use distillation on each other to avoid falling behind competitors. However, the practice remained largely unacknowledged in public statements and official channels. OpenAI, Anthropic, and Google have focused their public criticism on Chinese firms using distillation to create open-weight models available at much lower cost than U.S. offerings.

In February 2026, Anthropic accused several Chinese AI developers of using fraudulent accounts to extract large volumes of responses from Claude, its flagship chatbot, to train competing systems. The White House subsequently warned of "industrial-scale" campaigns using proxy accounts and jailbreaks to replicate U.S. AI capabilities. Yet Musk's testimony indicates the method is being used by U.S.-based companies as well, not only foreign competitors.

The irony is sharp. OpenAI and other frontier labs have themselves bent and allegedly broken copyright rules in their search for sufficient training data to build their models. Now they face a technique that turns their own work against them.

What Are the Legal and Ethical Boundaries?

The legal status of distillation remains murky. It is not explicitly illegal, but it can violate the terms of service that companies set for use of their products and APIs. OpenAI, Anthropic, and Google have reportedly launched an initiative through the Frontier Model Forum to share information about how to combat distillation attempts, typically by preventing users from making suspicious mass queries.

In August 2025, Anthropic took action by blocking OpenAI's access to Claude for violating the company's terms of service, which prohibit reverse-engineering services and building competing products. This suggests that while distillation may not be illegal, companies view it as a violation of their platform rules and competitive boundaries.

Musk's admission raises a critical question: if xAI, a company backed by one of the world's most influential entrepreneurs, is using distillation, how many other U.S. AI companies are doing the same? The practice appears to be far more widespread than the public record suggests.

Steps to Understand the Broader Implications of AI Distillation

• Recognize the Cost Advantage: Distillation allows smaller companies to achieve near-equivalent AI performance at a fraction of the computational cost, potentially democratizing access to advanced AI but also threatening the business models of companies that invested heavily in infrastructure.
• Understand the Terms of Service Risk: While distillation itself may not be illegal, companies using the technique expose themselves to legal action based on violations of platform terms of service, as demonstrated by Anthropic's action against OpenAI.
• Monitor Regulatory Developments: The legal boundaries around distillation remain unclear, and future court decisions or regulatory actions could establish whether the practice is permissible, making it important to track how Musk's lawsuit and similar cases unfold.
• Consider the Competitive Implications: If distillation becomes widespread and accepted, it could fundamentally reshape the AI industry by reducing the advantage of massive infrastructure investments and allowing more companies to compete at the frontier.

Why Did Musk Admit to This Practice Now?

Musk's testimony came during cross-examination by William Savitt, OpenAI's lawyer, who was pressing the question of whether xAI was truly a competitor to OpenAI or merely a tool to undermine the company. When asked directly if xAI used distillation techniques on OpenAI models, Musk responded "Partly," drawing audible gasps from the courtroom.

The admission appears to have been forced by direct questioning rather than volunteered. Savitt was building a case that Musk's lawsuit against OpenAI is motivated by competitive animus rather than genuine concern for the company's nonprofit mission. By revealing that xAI relies on OpenAI's technology, Musk inadvertently strengthened the argument that his company is indeed a direct competitor.

Musk's broader testimony painted a picture of a founder who felt deceived by OpenAI's shift from nonprofit to for-profit structure. He claimed he provided $38 million in essentially free funding to create what would become an $800 billion company, and that he only lost trust in CEO Sam Altman in late 2022 when he learned Microsoft would invest $10 billion in OpenAI. Yet the distillation admission complicates his narrative of principled opposition to OpenAI's commercialization.

What Does This Mean for the Future of AI Competition?

Musk's admission has broader implications for how AI companies will compete going forward. If distillation is standard practice among U.S. labs, as Musk claimed, then the competitive landscape is far more fluid than the public has understood. Companies can potentially leapfrog years of development by learning from their competitors' publicly available models.

This could accelerate AI progress by reducing barriers to entry for new competitors. However, it also threatens the business models of companies that have invested heavily in computing infrastructure and data acquisition. OpenAI, which has raised over $10 billion and spent billions on compute, faces the prospect that its investments can be partially replicated by competitors at a fraction of the cost.

The outcome of Musk's lawsuit against OpenAI could influence how courts and regulators view distillation. If Musk wins and forces OpenAI to restructure as a nonprofit, it could reshape the entire AI industry. Conversely, if OpenAI prevails, it may embolden other companies to pursue similar competitive strategies.

Meanwhile, xAI is expected to go public as part of Musk's rocket company SpaceX as early as June 2026, at a target valuation of $1.75 trillion, according to Musk's testimony. The company's reliance on distillation from OpenAI models raises questions about the sustainability of that valuation if the practice becomes legally or commercially restricted.