Trump veut interdire aux États américains de réguler l’IA pendant une décennie

Trump Proposes Moratorium on State AI Regulations for a Decade

As part of a comprehensive budget bill, a provision introduced by the Trump administration proposes prohibiting U.S. states from enacting or enforcing any AI-related regulations for ten years. This proposal has raised concerns within the technology sector and American civil society. The measure, currently under congressional debate, would prevent states from regulating high-risk AI systems as local legislation begins to address these issues.

Federal Moratorium with Extensive Implications

The provision is included in a legislative agenda that combines various fiscal and economic measures. If enacted, it would restrict states from regulating AI models, automated decision systems, or related technologies for a decade, leaving Congress as the sole regulatory body. This comes at a time when several states have started to legislate AI usage, targeting potential harms such as algorithmic discrimination in hiring, biases in healthcare, or misinformation through deepfakes.

A coalition of 141 organizations , including research institutions, civil society, and labor groups, have signed an open letter to Congress opposing the proposed moratorium. Notable signatories include Cornell University , the Center on Privacy and Technology at Georgetown Law , the Southern Poverty Law Center , the Economic Policy Institute , and employee groups like Amazon Employees for Climate Justice and the Alphabet Workers Union .

Currently, some U.S. states have enacted specific laws to regulate sensitive AI usage. For instance, Colorado requires companies to inform users when interacting with AI and prohibits algorithm-based discrimination. New Jersey has passed laws against the dissemination of misleading AI-generated content, particularly deepfakes.

States such as Ohio are considering transparency mandates through digital watermarks or penalties for AI identity theft. Many jurisdictions have also legislated on synthetic content use during elections.

Divergent Views Within the Tech Industry

The proposal contrasts with calls for regulation from some technology leaders. Sam Altman , CEO of OpenAI, emphasized the need for government regulation to mitigate risks from increasingly powerful models in a 2023 U.S. Senate hearing. He advocates for risk-based approaches with clear guidelines to navigate varying local regulations.

Other industry members, including those from Microsoft and Anthropic, share similar concerns that the absence of centralized regulation, combined with the prohibition of local initiatives, could erode public trust and fail to provide legal security for AI developers.

The Trump administration justifies this approach as essential to maintaining U.S. leadership in the global AI race, particularly against China. Vice President JD Vance argues that excessive regulation could hinder the burgeoning industry. This stance aligns with the administration's broader liberalization strategy, which includes lifting restrictions on AI chip exports to certain countries.

The proposed legislation must still pass multiple hurdles, having cleared a budget committee vote in the House and awaiting full House approval before Senate consideration.

D’après FrenchWeb.