ImageRights stands with the Digital Media Licensing Association (DMLA) in its formal response to the U.S. Office of Science and Technology Policy (OSTP) Request for Information on Regulatory Reform on Artificial Intelligence (Federal Register notice, Docket No. OSTP-TECH-2025-0067). The RFI asked the public to identify federal rules that hinder AI development, deployment, and adoption. DMLA's answer is direct: copyright is not the hindrance. Undermining it would be.

ImageRights founder and CEO Joe Naylor led the drafting of the response in his roles as President of DMLA and Chair of DMLA's Copyright & AI Working Group, and in his capacity representing DMLA within the Copyright Alliance, where coordinated advocacy across the creative industries informs and strengthens these aligned positions.

Why This Matters

The AI Action Plan released by the White House in July 2025 recognized that "high-quality data has become a national strategic asset." That data is overwhelmingly human-created copyrighted work. DMLA's submission lays out three principles for an AI industry that meets the standards of responsible, lawful, and ethical development:

1. It values and respects the rights of creators and copyright owners.
2. It does not make those rights subservient to the interests of AI developers.
3. It recognizes the value of the established and expanding licensing market for AI-related uses.

ImageRights supports DMLA on the core regulatory points the submission advances:

• Copyright is the foundation, not the barrier. With over fifty ongoing federal lawsuits on AI training, federal courts are already doing exactly what fair use was designed to let them do: apply a flexible, case-by-case framework to a new technology. No GAI exception in copyright law is needed, and creating one would destroy the incentive to produce the very works AI systems depend on.
• Free-market licensing is already working. Licensing initiatives from Getty Images, Shutterstock, Created by Humans, the Dataset Providers Alliance, Copyright Clearance Center, Elsevier, the Associated Press, and dozens of others demonstrate that the market for AI training rights is mature and growing. The Copyright Alliance maintains a public list of these agreements. The October 2025 settlement in Bartz v. Anthropic, which involved Anthropic's use of pirated works for training, reinforced that legitimate sourcing through licensing is the path forward, not the obstacle.
• Apply transparency to ingestion, not just outputs. Developers of publicly available GAI models that ingest unlicensed copyrighted works should be required to maintain and disclose adequate records of those works. The same standard should apply to retrieval-augmented generation (RAG) systems that source from copyrighted material at inference time. Transparency benefits creators and developers alike by building consumer and enterprise trust.
• Reject opt-out regimes as barriers to innovation. U.S. copyright law is unequivocally an opt-in framework, and that framework is the driving force behind the licensing market that already exists. Opt-out schemes, including the one adopted by the EU, are technically unworkable, ignore works already ingested, do nothing about pirated copies in third-party datasets, and would give AI developers an incentive to wait out the licensing market rather than participate in it.
• Prevent model collapse. Peer-reviewed research published in Nature in 2024 documented that AI models trained recursively on synthetic outputs degrade over time, forgetting the underlying distribution of real human work. Protecting copyright incentives keeps the supply of authentic, high-quality human-created training data flowing, which is what every future generation of GAI models will require to remain useful.
• Global competitiveness depends on IP strength, not its erosion. Robust copyright protection is what differentiates the U.S. from competitors that build AI capability by circumventing creator rights. Weakening copyright at home would surrender that advantage.

Our Commitment

As a long-time member and supporter of DMLA, and through DMLA's participation in the Copyright Alliance, ImageRights remains dedicated to providing the enforcement, registration, and monitoring tools creators need to operate in the generative AI era. The creators we serve produced the visual record that AI systems are now built on. Defending their rights is not a brake on innovation. It is the precondition for innovation that lasts.

Read the full DMLA response to OSTP on Regulatory Reform and join us in advocating for AI policy that recognizes copyright as a national strategic asset in its own right.