Skip to content
worth noting Regulation and law

Study: AI companies publicly support regulation, in practice they weaken it

only one source so far

A report by the AI Now Institute claims that companies developing frontier-generation models declare support for AI regulation, but in practice lobby against specific laws — both regarding the EU AI Act and California's SB-1047.

According to the report "Artificial Power: 2025 Landscape Report" by the AI Now Institute, companies developing frontier-generation models publicly declare support for AI regulation, but according to the authors, their actual behavior contradicts this. The report cites three cases as evidence: industry lobbying against rules for general-purpose models within the EU AI Act, the veto of the California SB-1047 law (Safe and Secure Innovation for Frontier Artificial Intelligence Models Act) despite its weakened final version, and federal efforts in the US to ban state regulation of AI.

The authors claim that competitive rivalry among a small number of frontier labs causes safety to take a back seat to rapid deployment and mass rollout of products. According to the report, political changes — the arrival of the second Trump administration in the US and a shift in sentiment in the EU — led a number of companies to abandon even their nominal support for regulation, which according to the authors suggests that these positions were tactical rather than genuinely motivated by public interest from the start.

The report also notes that support for regulation by itself is not an ideal indicator of behavior serving the public good: companies may support lax rules merely to fend off pressure for stronger laws, and poorly designed regulation can instead entrench companies' power and discretion under the guise of public legitimacy.

The source text is missing a further part of the analysis, including a detailed breakdown of individual companies' positions.

What changed

Why it matters

The report describes a pattern in which companies' public statements of support for AI regulation, according to the authors, do not match their lobbying and political behavior. This is relevant for assessing how much reliance can be placed on companies' nominal commitments to "safe AI" when evaluating regulatory risk and when tracking the development of specific laws such as the EU AI Act or SB-1047.

Relevant practical impact

What this means

01

For a business

Companies dealing with compliance for AI models should track actual legislative developments (EU AI Act, the status of SB-1047, federal efforts to ban state AI regulation in the US) instead of relying on declared support for regulation from AI model vendors, because according to the report, actual lobbying behavior differs from public statements.

Risks and compliance
What to decide Monitor the current status of the EU AI Act and any state/federal AI regulations in the US relevant to the models and vendors being used.
More business impacts →
AI security legislation lobbying corporate stance regulation

Check the original

Event sources

only one source so far · 1 publisher, 1 independent. We count feeds from the same owner only once.

1
AI Now Institute research source · first detected AI Regulation