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.
This month, ARL joined an amicus brief by the Consumer Technology Association (CTA) in support of fair use in the consolidated copyrighted cases against OpenAI. In September, the US Court of Appeals for the Third Circuit found that ROSS Intelligence’s copying of copyrighted legal content owned by Thomson Reuters to train a competing AI legal research tool is not fair use. In a spate of litigation, website and content owners allege that scraping their content to train generative AI tools is a violation of the Digital Millennium Copyright Act (DMCA).
The Re:Create Coalition has pushed back on recently introduced site-blocking bills; the bills exclude libraries and universities from site-blocking obligations, but raise broader concerns about online censorship.
And, the ARL Executive Committee approved an updated transparency statement in the context of AI; the statement explains that signing nondisclosure agreements (NDAs) with publishers can prohibit libraries from exposing patterns and building support for better negotiated terms.
The US Copyright Office announced its intention to recommend all DMCA-renewal petitions except for the current medical device repair exemption. In discussing the text-and-data mining exemption for literary works, the office acknowledged limitations of commercially licensed text-and-data mining products.
In Canada, three federal research funding agencies—the Canadian Institutes of Health Research (CIHR), the Natural Sciences and Engineering Research Council of Canada (NSERC), and the Social Sciences and Humanities Research Council of Canada (SSHRC)—announced their long-awaited revised Tri-Agency Open Access Policy on Publications.
CARL referred the House of Commons Standing Committee on Science and Research (SRSR) to its 2025 CARL “Brief on the Support for the Creation and Dissemination of Scientific Information in French,” and responded to the Government of Canada’s public consultation on advancing AI transparency in Canada.
Read on for more details, including CARL’s analysis of Bill C-34, the Safe Social Media Act, and Bill C-36, an Act to Enact the Protecting Privacy and Consumer Data Act.
|
AI Transparency in Canada
|
CARL Analysis of Federal Digital Policy Bills C-34 & C-36
|
Copyright and Fair Use/Fair Dealing
|
AI Transparency in Canada
|
CARL Responds to Federal Survey on AI Transparency
|
CARL responded to the Government of Canada’s public consultation on advancing AI transparency in Canada. The survey was initiated by Innovation, Science, and Economic Development and contributes to Canada’s new AI strategy, AI for All, which was launched in June 2026. The government aims to advance AI transparency so Canadians can better understand when and how they interact with AI. While not addressing all AI-related risks, greater transparency can support informed decision-making, accountability, responsible business practices, and identify where further government intervention is needed.
|
Tri-Agencies Announce Open Access Policy on Publications
|
Canada’s three federal research funding agencies—the Canadian Institutes of Health Research (CIHR), the Natural Sciences and Engineering Research Council of Canada (NSERC), and the Social Sciences and Humanities Research Council of Canada (SSHRC)—announced their long-awaited revised Tri-Agency Open Access Policy on Publications (2026). The revised policy requires all peer-reviewed research articles from federally funded research grants to be immediately and freely available on publication and will apply to all grants and chairholder awards, including those administered by the Tri-agency Institutional Programs Secretariat (TIPS), awarded on or after January 1, 2027. The policy requires peer-reviewed research articles to be deposited in a Canadian institutional repository or made available through a trusted Canadian platform at the time of publication, which acknowledges that research libraries are indelible partners in the research ecosystem.
|
CARL Highlights Libraries’ Role in French-Language Science
|
CARL Analysis of Federal Digital Policy Bills C-34 & C-36
|
Bill C-34, the Safe Social Media Act
|
If passed, Bill C-34 will introduce a new online-safety regime, including obligations for social media services and chatbots, with requirements only applying to agentic chatbots and those that simulate human interaction. The bill is primarily intended to protect children, address the spread of harmful online content, and increase platform accountability. CARL’s current understanding is that this bill is unlikely to directly impact libraries. Although some libraries own their chatbot services, CARL recognizes that this would be an institutional responsibility related to the adoption of agentic chatbots and has determined this is outside CARL’s current purview.
|
Bill C-36, An Act to Enact the Protecting Privacy and Consumer Data Act
|
If enacted, Bill 36 would replace PIPEDA as the primary Canadian federal privacy statute governing the collection, use, and disclosure of personal information in the course of commercial activities, including by federally regulated employers in relation to employee personal information. Since this bill’s focus is on commercial activities and universities are not subject to PIPEDA, CARL does not believe that this legislation will have any direct impact on research libraries if it passes. With this understanding, CARL will not respond or submit feedback.
|
Copyright and Fair Use/Fair Dealing
|
US Register of Copyrights Intends to Recommend Renewing Text-and-Data Mining Exemption, Acknowledges License Limitations
|
At the beginning of this Digital Millennium Copyright Act (DMCA) rulemaking cycle, Authors Alliance and the Library Copyright Alliance (LCA) petitioned to renew the text-and-data mining (TDM) exemption for literary works and motion pictures. The exemption permits researchers to circumvent digital locks in order to assemble and annotate their own research corpora of copyrighted works, and has enabled individual research projects while contributing to the development of new research methods.
In September, the Association of American Publishers (AAP) filed an opposition to the literary works petition, arguing that it is no longer necessary because licensing and market solutions for TDM research continue to emerge and expand, citing examples like ProQuest TDM Studio.
In October, the US Copyright Office stated the intention of the register of copyrights to recommend renewing the petition, despite the AAP opposition. The office concluded that the conditions that led to the adoption of the exemption are likely to continue:
Contrary to AAP’s assertion, Authors Alliance and LCA state in their petition that “[c]ommercially licensed [TDM] products continue to be made available to research institutions” but “these licensed products do not allow researchers to license the full array of texts needed to engage in the research they seek to do.”
The Copyright Office also announced the register’s intention to recommend all DMCA renewal petitions except for the current medical device repair exemption.
ARL welcomes the Copyright Office’s recognition that licensed products do not always meet the full needs of TDM researchers. Renewing the exemption will preserve researchers’ ability to build corpora suited to their questions, and will allow the important work enabled by the exemption to continue.
|
Library Copyright Alliance Joins Amicus Brief in Support of Fair Use
|
The Association of Research Libraries (ARL) and American Library Association (ALA)—together, the Library Copyright Alliance (LCA)—joined an amicus brief in favor of fair use, filed on October 30 in the consolidated copyright cases against OpenAI and Microsoft, including the New York Times’s suits. This is among the first AI copyright cases to reach the summary judgment phase, and its outcome is likely to influence how courts treat fair use across the dozens of similar cases now pending.
LCA’s interest is in preserving the fair use doctrine to support current and future AI applications for research and education. Libraries spend over $1 billion annually on subscriptions for published scholarly literature that researchers in science, medicine, and the humanities use to train AI models and conduct text-and-data mining research. Narrowing fair use would limit how researchers can use this material; leaving fair use intact gives libraries and researchers the flexibility to freely pursue important research questions without seeking permission for every computational use.
LCA’s participation in this brief flows from LCA’s statement on copyright and generative AI, which posits that ingesting lawfully accessed copyrighted works to train large language models (LLMs) is generally a fair use under US copyright law. The bedrock of LCA’s analysis is the Google Books litigation, in which the US Court of Appeals for the Second Circuit held that copying to build a searchable database is highly transformative—a key consideration in the first factor of a fair use analysis. According to LCA’s analysis, the same logic applies to training an LLM. The Second Circuit also found no market harm from the copying in Google Books, since users could not access the full text of the copied books. This is critical to the fourth factor, which asks whether a copy substitutes for the original expression in the same market.
The Google Books case is especially relevant because any appeal in New York Times v. OpenAI and Microsoft would go before the Second Circuit, which likely would follow its own precedent.
Courts have not treated commerciality as decisive in the Google Books cases, or in the other AI cases in which training was determined to be fair use (Bartz v. Anthropic and Kadrey v. Meta). This is notable because it means researchers, scholars, libraries, and commercial entities all rely on the same fair use right to enable innovation and access to knowledge.
The amicus brief, led by the Consumer Technology Association (CTA), includes perspectives on fair use and accessibility for people with disabilities, education, research, journalism, and access to information. Other amici include the American Council for the Blind and the Business Software Alliance.
|
ARL Executive Committee Approves Updated Statement on Transparency in Licenses for Electronic Resources
|
ARL’s Executive Committee approved an amended statement on transparency as a core operating principle of licensing, in light of new considerations around AI. Developed in consultation with ARL’s Advocacy and Public Policy Committee, the new sections address terms that restrict the use of licensed electronic resources with AI:
Research libraries license content to support the full range of teaching, learning, and research activity, including advanced computational research such as text-and-data mining, generative AI, non-generative AI analysis, as well as foundational data practices. Increasingly, however, some vendors are inserting blanket clauses into license agreements that broadly restrict the use, creation, or input of licensed content into AI tools. These restrictions often extend beyond AI-specific applications to uses within everyday productivity and office software, such as word processors, citation managers, and email, which now have AI-embedded features by default. In some cases, the language is broad enough to prohibit use altogether, given how AI capabilities are present in standard academic and research platforms. Such clauses are unworkable because they unduly restrict teaching and research functions with no regard for a user’s intent, the purpose of the use, or the pervasiveness of AI tools across common educational and research platforms. License agreements should enable research activity, not constrain it, and a license that technically prohibits “AI use” today may be prohibiting normal scholarly work tomorrow.
Declining or narrowing NDA terms means libraries can share information about institutional license terms with faculty, administrators, and researchers within the institution. That transparency also extends towards other libraries and peer institutions, where it can also expose patterns and build support for collective action for better negotiated terms.
|
Re:Create Coalition Opposes Site-Blocking Bills
|
Bipartisan members of Congress have introduced site-blocking legislation that would establish procedures for rightholders to petition a district court to designate certain websites as “foreign piracy sites.” The court could then order the designated sites to be blocked for users in the US.
The bills exclude libraries and universities from site-blocking obligations. However, as Re:Create explained, similar bills in European countries have caused collateral damage by disrupting access to legitimate websites, including those hosted by universities and libraries. (ARL is a member of Re:Create.) In October 2026, the privacy-focused DNS provider Quad9 announced that it will cease all operations in France following a French court ruling that ordered the nonprofit to block certain streaming domains.
Rightsholders in the US already have several options for recourse for alleged infringement, including issuing a takedown notice or subpoena under the Digital Millennium Copyright Act (DMCA), or seeking an injunction in federal court.
Congress has tried and failed to pass site-blocking legislation since the introduction of the Stop Online Piracy Act (SOPA) in the House and the PROTECT IP Act (PIPA) in the Senate; votes on SOPA/PIPA were indefinitely postponed in 2012 due to a massive anti-censorship campaign organized by digital rights advocates.
The House Judiciary Committee may mark up the bills, which include the American Copyright Protection Act of 2026, sponsored by Rep. Issa (R-CA), and the DEFEND IP Act, introduced by Rep. Lofgren (D-CA) and Rep. Cline (R-VA). Even if these bills advance out of committee, it is unlikely that site-blocking legislation will become law this year. ARL will continue to track developments and share updates.
|
Website Owners Allege Web Scraping Violates DMCA Anti-Circumvention Provisions
|
In a spate of more than 20 lawsuits, website publishers allege that scraping online content to train generative AI models despite the presence of tools like YouTube’s “rolling cipher,” Google’s SearchGuard, CAPTCHA, and robots.txt violates the Digital Millennium Copyright Act (DMCA). The DMCA prohibits circumventing technological protection measures (TPMs) that effectively control access to copyrighted works, and trafficking in technologies that enable circumvention (Sections 1201(a)(1)(A) and (a)(2)).
Key issues in the cases include whether such tools effectively control access to a copyrighted work, if the TPM controls access rather than copying, and whether a defendant circumvented an effective control (some courts have described an effective control as one that prevents all methods of access). The Southern District of New York has already rejected a claim based on robots.txt, holding that the files are requests rather than technological barriers and that ignoring them is not circumvention of an effective technological measure (Ziff Davis v. OpenAI).
ARL will monitor this litigation for potential implications for web scraping by researchers.
|
Appellate Court Finds ROSS’s Copyright of Thomson Reuters Headnotes Is Not Fair Use
|
An appellate court affirmed a district court decision that Thomson Reuters’s materials are copyrightable, and that ROSS’s use of them to create a competing AI legal research platform was not fair.
The case began when Thomson Reuters sued ROSS Intelligence, alleging that ROSS training its AI legal research platform on Westlaw content constitutes copyright infringement (Thomson Reuters owns West publishing and copyright in its Westlaw content). The content in question was Westlaw’s headnotes, which describe points of law in a judicial decision.
In February 2025, a district court judge ruled in favor of Thomson Reuters and West, finding that the headnotes were copyrightable and that ROSS’s use did not qualify as fair use. The judge emphasized the fourth fair use factor—market harm—noting that ROSS developed its tool specifically to compete with Westlaw.
In September 2026, the US Court of Appeals for the Third Circuit affirmed the district court’s decision that the headnotes were protected expression and that ROSS’s use was not fair. While the factual nature of the published headnotes weighed in favor of the second fair use factor, the court ultimately found that the first, third, and fourth factors weighed against it. In its analysis of the fourth factor, which weighs whether a copy creates a market substitute, the court noted that ROSS did not hide its intention that this for-profit AI platform compete with Westlaw.
The court was careful to note that while the case appears to concern the future of AI legal technology, in truth it is no more than an ordinary copyright case. The court also noted that ROSS’s AI was not generative AI, in that it did not create any new expression.
The Library Copyright Alliance joined an amicus brief in the case, urging the appellate court to reverse the district court’s adverse fair use ruling.
|
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.
|
| |
|
|
