Katherine Klosek, Director, Information Policy and Federal Relations, ARL
Contributions from the Canadian Association of Research Libraries (CARL) by Katherine McColgan, Manager, Administration and Programs, CARL
|
The monthly Public Policy Briefing highlights developments in ARL’s public policy priorities, issues that are relevant to the research library community in the United States and Canada, and details on advocacy conducted by ARL and CARL. Please encourage your colleagues to sign up for the Public Policy Briefing.
If you have questions or suggestions, please email me at kklosek@arl.org.
In July, the Library of Congress deputy CIO testified in a congressional hearing about efforts to continue treating machines as distinct users of its most popular websites. US Office of Science and Technology Policy (OSTP) Director Michael Kratsios testified in favor of legislation to regulate the use of unauthorized deepfakes; he reinforced the Trump administration’s position that “publicly available materials should be allowed for the training of models.” ARL is analyzing bills that would regulate chatbots available to minors with a focus on implications for libraries.
ARL continues to monitor the consolidated copyright cases against OpenAI. A judge approved the settlement agreement in Bartz v. Anthropic. A court in India held that OpenAI training its models on copyrighted works is fair dealing. ARL joined 26 other associations in an amicus brief in support of Harvard in its grant-funding litigation against the Trump administration.
The White House signaled its vision for a shift in the traditional role of research libraries in federally funded research through a joint OSTP/Office of Management and Budget (OMB) memo and an OSTP report. The Senate included language blocking OMB’s proposed overhaul of the Uniform Guidance in its stopgap spending proposal.
CARL and the Canadian Research Knowledge Network (CRKN) affirmed their shared commitments to open science in a joint statement.
Read on for more details!
|
Copyright and Fair Use/Fair Dealing
|
Federally Funded Research
|
Library of Congress Deputy CIO Testifies on Treating Machines as Distinct Users
|
During a hearing of the US Committee on House Administration titled “Modernizing Public Access to Legislative Data,” Library of Congress Deputy Chief Information Officer John Rutledge testified about the library’s efforts to enhance the user experience of Congress.gov and LOC.gov—the library’s two most popular websites—for both humans and machines.
According to Rutledge, the Congress.gov API has seen growth in requests coming from AI crawlers and assistants. This recent increase signals that the information available on Congress.gov is no longer limited to the site itself; it now appears in AI-assisted search results as well as web, mobile, and desktop AI applications. The library intends to treat this machine access as a distinct class of users, and to factor their methods and needs into information platform updates.
Separate from the API traffic, Congress.gov has seen a recent spike in AI-training crawlers designed to access sites through the same interfaces as human users. This extraordinarily high volume consumed system resources and degraded performance for public users. Implementing a human verification check on the search page significantly improved the site’s performance and blocked approximately 98% of incoming requests as nonhuman traffic.
To meet increasing machine-traffic demands, the library’s FY 2027 appropriations request includes funding for an AI enterprise platform and moving Congress.gov to cloud hosting.
|
ARL Analyzing Federal AI Chatbot Bills for Potential Library Liability
|
Two bills have been introduced in the US Senate that would impose restrictions on providers of AI chatbots to minors. Although neither bill specifically targets research libraries, their scope is sufficiently broad that they might apply to research libraries that make chatbots available to students under the age of 18.
The CHATBOT Act, a bipartisan bill led by Senator Cruz (R-TX), would require AI chat providers—which could include libraries—to allow parents or guardians the option to access a full record of their teens’ conversations. If a library develops, hosts, or deploys its own proprietary AI chatbot that serves minors, it could be covered under the CHATBOT Act. Covered entities that violate the bill would be subject to enforcement action by the Federal Trade Commission (FTC) or a state attorney general.
Under the Youth AI Privacy Act, introduced by Senator Markey (D-MA), AI chatbot providers would be required to disclose to users under age 18 that they are not interacting with a human, and that the chatbot’s content is AI-generated. The FTC would have discretion to determine whether these disclosures are in compliance with the bill’s requirements. The bill also bars processing a minor’s personal data for profiling and prohibits processing or transferring their data to train AI models. Crucially, a private right of action would allow a parent or guardian to sue libraries directly for a violation of these provisions.
While neither bill is likely to pass in this session, ARL is monitoring the legislative trend toward holding institutions responsible for the AI tools they host. ARL will engage with lawmakers to ensure that regulations don’t unduly introduce liability for research libraries or restrict access to information.
|
Copyright and Fair Use/Fair Dealing
|
Amicus Briefs Due October 15 in OpenAI Consolidated Cases
|
In September, OpenAI will file a motion for summary judgment arguing that training its models on copyrighted works is fair use. The case consolidates 18 lawsuits alleging infringement through OpenAI’s use of books and news content, which are currently pending in the Southern District of New York.
OpenAI’s motion is expected to align with the fair use arguments that were successful in the Anthropic and Meta cases. A central argument will be that OpenAI’s use transformed the copyrighted works into something new with a different purpose and character, rather than reproducing the training data.
ARL is closely monitoring this litigation. A decision that ingestion for training purposes falls outside fair use could have a negative impact on AI training in other contexts, such as AI-powered research. Amicus briefs are due on October 15, 2026.
|
Judge Approves Settlement in Bartz v. Anthropic
|
Thousands of authors and publishers who were part of the class action copyright suit Bartz v. Anthropic will split $3,000 per book, according to a settlement agreement approved by a federal judge in the US District Court for the Northern District of California.
In August 2025, the Library Copyright Alliance (LCA) joined Authors Alliance, Electronic Frontier Foundation (EFF), American Library Association (ALA), and Public Knowledge in an amicus brief asking the US Court of Appeals for the Ninth Circuit to reverse the class certification. Amici argued that class certification would force Anthropic to settle, leaving important copyright questions unresolved by the courts.
Anthropic was scheduled to go to trial on the issue of acquiring books through pirate websites, but that trial was canceled when the company agreed to the settlement. There is no final judgment on the question of whether using unauthorized copies affects a fair use analysis.
|
OSTP Director Michael Krastios Addresses NO FAKES Act, Reaffirms AI Training Stance Before Congress
|
During a US House of Representatives hearing on “Unleashing the Golden Age of Science: Examining the Priorities of the FY 2027 Research and Technology Enterprise," Office of Science and Technology Policy (OSTP) Director Michael Kratsios said that the White House priorities on AI include the topic of protecting image and likeness. The comment came in response to questioning from Rep. Friedman (D-CA), a co-sponsor of the NO FAKES Act, who asked whether Kratsios would commit to working with Congress on the bill.
The NO FAKES Act would create liability for unauthorized use of a person’s image or likeness, though it exempts libraries, archives, and educational institutions under language the Library Copyright Alliance negotiated into the bill.
In the same line of questioning, Kratsios reaffirmed the Trump administration’s position that publicly available materials should be permitted for training AI models.
See the latest edition of the Monitor for more coverage of the hearing.
|
Court in India Permits AI Training on Copyrighted Works
|
The Delhi High Court ruled that OpenAI training its large language models (LLMs) on copyrighted works constitutes fair dealing.
The case concerns news agency ANI Media Private Ltd., which raised issues of infringement of its copyrighted material for LLM training by OpenAI in 2024. Relying on the fair use findings in Bartz v. Anthropic and Kadrey v. Meta, the court concluded that OpenAI’s act of storing ANI’s original retrieved works for training LLMs underlying ChatGPT is fair dealing under Indian copyright law. Further, ANI failed to show the court that “any memorization or regurgitation of [its] original retrieved works has happened through the responses generated by ChatGPT.” The court also held that ChatGPT’s retrieval-augmented generation (RAG) outputs were not substantially similar to ANI’s original content, and therefore did not amount to infringement.
|
Join ARL and CARL for Charleston Preconference Workshop: The Right to Research in the Age of AI
|
The Association of Research Libraries (ARL) and the Canadian Association of Research Libraries (CARL) are hosting a workshop before the 2026 Charleston Library Conference on Tuesday, November 3, to define conditions for advanced computational research, identify red-flag license terms, and develop shared negotiation strategies, including approaches for engaging faculty and administrators. We hope you will register and join the conversation!
Recent conversations at the ARL Spring 2026 Association Meeting and a joint CARL/ARL presentation at the 2026 ABC Copyright Conference surfaced a challenge and an opportunity. Researchers using licensed databases and journals likely never read the license terms governing the use of that content, but these terms have major implications for their right to download or otherwise access content; engage in computational analysis like text and data mining; and retain data for purposes of replication and validation of their results. Failure to reach timely agreement on terms can lead to interim loss of access to collections & resources. In a more extreme example, a paper was retracted when publishers asserted that researchers’ methods did not comply with the terms of their institution’s license agreement.
In recent years, publishers have introduced blanket clauses restricting the usage of AI with “any AI tool.” But because many AI tools are automatically embedded in vendor platforms, this puts librarians in a difficult position: sign a contract they know the institution will immediately and automatically violate, or walk away from the negotiation and discontinue access, disrupting the research operations of the university. This also erodes the hard-won progress that libraries have made in securing text and data mining (TDM) provisions in license agreements or amendments, since many TDM methodologies rely on non-generative AI systems to extract information.
The November 3 workshop is a space to move from identifying the problem to building collective solutions.
|
Federally Funded Research
|
CARL and CRKN Affirm Ongoing Shared Commitments to Open Science in Joint Statement
|
CARL and the Canadian Research Knowledge Network (CRKN) announced a recommitment to work together to lay the foundation for a renewed vision for collaborating to expand open science and drive the evolution of open knowledge in Canada. Begun in 2023, the two organizations will continue to coordinate efforts around key areas, such as the new Tri-Agency Open Access Policy on Publications and digital research infrastructure.
|
White House Signals Shift in Research Enterprise
|
In the annual memo on White House FY 2028 R&D priorities, the Office of Science and Technology Policy (OSTP) and Office of Management and Budget (OMB) laid out the Trump administration’s research priorities for FY 2028, as well as a set of “R&D Priority Practices” with a shorter timeline for implementation using existing resources.
Both sets of priorities have potential implications for research libraries’ role in the research enterprise. For instance, the administration directs agencies to invest and prioritize incentives for researchers to make scientific datasets and data infrastructure accessible for AI training in their FY 2028 budget submissions to OMB. Meanwhile, agencies must align existing resources to meet science-of-science priorities, which include requirements for agencies to evaluate how factors such as funding mechanisms, peer review, and publication practices affect scientific outcomes.
The memo accompanied an OSTP report to the president, Science: A New Golden Age, which signals the administration’s vision to rethink universities’ traditional role in federally funded research. The report does not have the weight of a memo or executive order.
|
Senate Stopgap Funding Proposal Would Pause OMB Overhaul of Uniform Guidance
|
ARL Joins Amicus Brief in Support of Harvard
|
ARL joined 26 other associations, including the Association of American Universities (AAU), COGR, and EDUCAUSE, in an amicus brief by the American Council on Education (ACE) in support of Harvard in the case Harvard v. US Department of Health and Human Services, opposing the Trump administration’s efforts to punish Harvard by stripping it of grant funding and other retaliatory actions.
The case began in April 2025, when the administration froze more than $2.2 billion in research grants after Harvard declined to comply with its set of demands. Harvard sued, arguing the freeze was unlawful retaliation and violated its First Amendment rights. In September 2025, a District Court judge issued an injunction that halted the funding freeze and grant terminations. In the brief, amici urged the US Court of Appeals for the First Circuit to affirm the District Court’s injunction.
The brief reads:
The Administration’s actions in this case are directed at Harvard, but they reverberate throughout the nation. If the federal government may punish a university for the perceived ideology of its professors, then the marketplace of ideas that is a core part of American higher education will be transformed into a state-held monopoly on controlling academic philosophy. That is the antithesis of America’s constitutional values, and it jeopardizes the richness of the spectrum of higher education that has long been one of our country’s greatest strengths.
The administration’s appeal is currently pending.
|
About the Association of Research Libraries
|
The Association of Research Libraries (ARL) is a nonprofit membership organization of research libraries and archives in major public and private universities, federal government agencies, and large public institutions in Canada and the US. ARL champions research libraries and archives, develops visionary leaders, and shapes policy for the equitable advancement of knowledge.
|
|
|
|
Manage your preferences | Opt Out using TrueRemove™
Got this as a forward? Sign up to receive our future emails.
View this email online.
|
1025 Connecticut Avenue NW #1200 | Washington, DC 20036 US
|
|
|
|
This email was sent to .
To continue receiving our emails, add us to your address book.
|
| |
|
|
