Senator Elizabeth Warren is picking a fight with the AI industry, and this time the battlefield is a trade agreement most people forgot existed.

On July 23, Warren publicly accused major AI corporations of lobbying the Trump administration to water down transparency and oversight provisions in the updated US-Mexico-Canada Agreement. Her argument: Big Tech is trying to use international trade law as a shield against domestic regulation.

What Warren actually wants

In a letter dated July 22 and addressed to US Trade Representative Jamieson Greer, Warren laid out her core complaint. Under existing USMCA rules, AI companies are only required to disclose details about their models, like source code, algorithms, and training data, during formal government investigations.

She urged Greer to expand regulatory access to this information even without an active probe into a specific company. The senator characterized the current framework as a win for Big Tech, one that entrenches protections for proprietary AI systems and keeps regulators perpetually one step behind.

Warren didn’t name specific companies as lobbyists, instead pointing broadly at “major AI companies” and their industry groups. Trade associations representing Google, Meta, Apple, and Amazon Web Services have previously communicated their concerns to the USTR about the risks of mandated disclosures. Their argument: forcing companies to reveal source code, algorithms, model weights, and training data could jeopardize their AI services.

The broader context

Warren’s letter doesn’t exist in a vacuum. In February 2026, she sent a separate letter raising concerns about so-called “reverse acqui-hire” deals, transactions where large tech companies effectively absorb AI startups and their talent without triggering traditional merger review.

The AI industry has also faced a rough stretch of public trust issues. A lawsuit against OpenAI has added to the sense that these companies are moving fast and breaking things in ways that carry real consequences. A separate incident involving a rogue AI agent tied to Hugging Face further amplified concerns about the gap between AI capabilities and the oversight mechanisms meant to govern them.

For Warren, the USMCA renegotiation represents a strategic chokepoint. Trade agreements are notoriously sticky once finalized. Provisions baked into the updated accord could effectively lock in a permissive regulatory environment for AI companies for years, making it exponentially harder for future administrations or Congress to impose meaningful transparency requirements domestically.