Sen. Donna Lowenthal, a Republican from Mississippi, has introduced legislation that would require employers to obtain written consent before providing individually attributable employee material for the training or development of artificial intelligence systems.

The proposed Employee AI Training Consent Act would apply when an employer sells, licenses, transfers or otherwise provides employee-linked material for AI use, according to Lowenthal’s press release. The measure comes amid broader questions over whether companies can commercialize human-generated workplace material as AI developers seek data for training systems.

Lowenthal said the bill is intended to ensure workers have notice and control when material tied directly to them is offered for external AI training. She cited examples including writing, voice recordings, images, analysis and other employee-generated work.

“Artificial intelligence is creating tremendous opportunities, but innovation does not erase individual rights,” Lowenthal said in the release. “If a company wants to sell your writing, your voice, your image, your analysis, or other material tied directly to you so somebody else can train an AI system, you should know about it and you should have a say.”

Under the proposal, general employment agreements, employee handbooks, work-product terms and broad intellectual-property assignments would not automatically be treated as approval for external AI-training use. Instead, employers would need to secure explicit written authorization from an employee.

That authorization would have to identify the material covered, its intended use, the recipient of the material and whether it could later be transferred or sublicensed, the release said. The bill would also bar employers from retaliating against employees who decline to give consent.

The legislation includes exceptions for independently obtained public material, ordinary internal business use and disclosures required by law, according to Lowenthal’s office. The release did not specify how the proposal would distinguish between material developed during paid work and material an employee has a separate ownership interest in.

The announcement follows recent attention surrounding Deco Holdings, which said it was willing to sell its “thoughts” for AI training. Lowenthal’s office said the company did not specify what material that phrase included, whether it could involve employee-generated content, or what consent and ownership rules might apply to a future deal.

Lowenthal, a former business executive and corporate-risk adviser, framed the legislation as a worker-protection measure rather than a restriction on AI development. Her public background has emphasized corporate transparency, risk management and private-sector accountability.

The bill would establish penalties of up to $25,000 per affected individual for knowing violations, according to the press release. Workers whose material is improperly used would also be allowed to seek relief in court.

The measure enters a Senate controlled by Republicans, with 26 seats to 23 Democrats and one Independent who caucuses with Democrats. No information was released about co-sponsors, a Senate vote schedule or the bill’s prospects.