MAGA cognoscenti warn feds away from shielding AI infringers
(2025/12/03)
- Reference: 1764797864
- News link: https://www.theregister.co.uk/2025/12/03/maga_bannon_anti_ai_fair_use/
- Source link:
A group of conservatives allied with President Donald Trump's MAGA movement, including former Trump strategist Steve Bannon, has asked the Justice Department and the White House to stop protecting Big Tech against copyright claims.
In [1]a letter [PDF] addressed to US Attorney General Pam Bondi and Michael Kratsios, director of the White House Office of Science and Technology Policy, more than a dozen leaders of conservative organizations urge the government to reject [2]calls to change copyright law to accommodate AI companies.
The signatories argue that supporting the tech industry's effort to reform copyright law would harm American workers, contradict the Trump administration's AI and trade policy, dilute US soft power, and encourage economic espionage by China.
[3]
"Opponents of applying copyright law to AI make a series of flawed claims in an effort to cut corners," the letter says. "One is that competition with China requires the US to abandon its commitment to strong IP protections."
[4]
[5]
The authors contend that weakening IP rights domestically will allow China and other adversaries "to avail themselves of the same dubious 'fair use' theories to not only steal creative content, but also proprietary US AI models and algorithms."
Prior to the debut of OpenAI's ChatGPT in November 2022, legal scholars voiced support for the tech industry's view that training AI models on copyrighted content qualified for the fair use defense against infringement claims.
[6]
The [7]fair use defense under US law requires an analysis of four factors: the purpose and character of the use, the nature of the copyrighted work, the amount of material used, and the effect upon the market for the copyrighted work.
In 2021, for example, judicial law clerk Jenny Quang, who at the time was a UC Berkeley Law School research assistant, wrote an [8]article [PDF] for the Berkeley Technology Law Journal titled, "Does Training AI Violate Copyright Law?"
Quang wrote, "Training a machine learning model with this copyrighted data does not infringe because the data are not redistributed or recommunicated to the public. Copyright protects creative expression, but model training extracts unprotectable ideas and patterns from data."
[9]
That same year, Nat Friedman, then CEO of GitHub, echoed that line of reasoning as GitHub debuted a beta version of its Copilot AI assistant. He [10]wrote in a social media post , "training ML systems on public data is fair use [and] the output belongs to the operator, just like with a compiler."
[11]Amazon is forging a walled garden for enterprise AI
[12]Zig quits GitHub, says Microsoft's AI obsession has ruined the service
[13]China using AI as 'precision instrument' of censorship and repression, at home and abroad
[14]Apple swaps one ex-Google AI chief for another
Various legal scholars have made [15]similar [16]arguments .
But the AI-related copyright claims heard to date by US courts suggest AI companies can't count on a fair use defense in all cases – and there are [17]more than 50 of them presently.
When legal scholar Pamela Samuelson asked the same question in October in [18]an article titled "Does Using In-Copyright Works as Training Data Infringe?", she came to a less definitive conclusion based on the outcomes of [19]Bartz v. Anthropic and [20]Kadrey v. Meta .
In both cases, authors sued the AI providers (Anthropic and Meta) for training on copyrighted books, and while the judges mostly found for the AI companies, the results weren't completely cut and dried. In Anthropic, the judge ruled that as long as the company acquired the copyrighted material legally, it could train its AI on it; Anthropic later settled with authors of some books that it had pirated. In the Meta case, the judge said that it was fair to use the copyrighted texts and the plaintiffs did not present enough evidence of market harm, which at least paves the way for future rulings to take market effects into consideration.
"The Bartz and Kadrey decisions are mixed bags," she wrote. "They indicate that some training data uses may be found fair use, while others may not."
One of the things that has changed in the past few years is that the [21]market impact of AI on copyrighted works is starting to show up.
In Thomson Reuters' copyright claim against Ross Intelligence, the court in February found that training AI models on Westlaw content to build a rival legal service [22]did not qualify as fair use .
AI companies have recognized the risk of infringement claims by [23]settling or striking content [24]licensing [25]deals , though they'd prefer not to incur that cost.
And the potential cost is high because so much of what people have written is subject to copyright protection. As OpenAI CEO Sam Altman said in a December 2023 [26]submission [PDF] to an inquiry by the UK House of Lords, leading AI models could not be trained without the use of copyrighted content.
"Limiting training data to public domain books and drawings created more than a century ago might yield an interesting experiment, but would not provide AI systems that meet the needs of today's citizens," he said.
If AI companies were broadly required to pay for the data they use, their already extensive data center expenditures would balloon with the cost of licensing arrangements.
The conservative letter signatories say that's just fine.
"It is absurd to suggest that licensing copyrighted content is a financial hindrance to a $20 trillion industry spending hundreds of billions of dollars per year," the letter says. "AI companies enjoy virtually unlimited access to financing. In a free market, businesses pay for the inputs they need. Imagine if AI CEOs claimed they needed free access to semiconductors, energy, researchers, and developers to build their products. They would be laughed out of their boardrooms."
Adam Eisgrau, senior director of AI, Creativity and Copyright Policy at Chamber of Progress – one of the tech-aligned groups cited by the conservative authors – rejects the call to have the feds remain on the sidelines.
In an email to The Register , he said, "The purpose of copyright as defined in the Constitution is to promote innovation: what the Framers called the 'progress of science and useful arts.' Under the law, by definition fair use isn't theft or copyright infringement; it's exactly what the Framers wanted. The DOJ absolutely should go to bat for innovators and fair use."
The letter concludes, "President Trump – himself a bestselling author and former television producer – said it best: 'The pioneering spirit of [our] artists, authors, inventors, and other creators has improved our lives and the lives of millions of people around the world, and will continue to propel us toward a better future.'"
Trump made those remarks [27]in April 2020 [PDF], before [28]AI lobbying surged . We note that earlier this year, the President fired Shira Perlmutter, Register of Copyrights, an event that US Representative Joe Morelle (NY-25) [29]linked to Perlmutter's refusal "to rubber-stamp Elon Musk's efforts to mine troves of copyrighted works to train AI models." ®
Get our [30]Tech Resources
[1] https://cdn.prod.website-files.com/65a5dbf7ffda034a47a316a2/692de157f04ef0ddcecc1d34_Coalition%20Letter%20to%20DoJ%20and%20OSTP%20re%20AI%20and%20Copyright_12.1.25.pdf
[2] https://progresschamber.org/resources/white-house-request-for-information-regulatory-reform-on-artificial-intelligencefederal-register-docket-no-ostp-2025-0005/
[3] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2aTDBBznNocGx8l5NdhdgfQAAAMA&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[4] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aTDBBznNocGx8l5NdhdgfQAAAMA&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[5] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33aTDBBznNocGx8l5NdhdgfQAAAMA&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[6] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aTDBBznNocGx8l5NdhdgfQAAAMA&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[7] https://www.copyright.gov/fair-use/
[8] https://btlj.org/wp-content/uploads/2023/02/0003-36-4Quang.pdf
[9] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33aTDBBznNocGx8l5NdhdgfQAAAMA&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[10] https://twitter.com/natfriedman/status/1409914420579344385
[11] https://www.theregister.com/2025/12/03/amazon_enterprise_ai_walled_garden/
[12] https://www.theregister.com/2025/12/02/zig_quits_github_microsoft_ai_obsession/
[13] https://www.theregister.com/2025/12/03/aspi_china_ai_report/
[14] https://www.theregister.com/2025/12/02/apple_ai_chief/
[15] https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4986763
[16] https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5395242
[17] https://chatgptiseatingtheworld.com/2025/10/08/status-of-all-51-copyright-lawsuits-v-ai-oct-8-2025-no-more-decisions-on-fair-use-in-2025/
[18] https://cacm.acm.org/opinion/does-using-in-copyright-works-as-training-data-infringe/
[19] https://www.theregister.com/2025/09/08/anthropic_settles_author_lawsuit/
[20] https://www.theregister.com/2025/06/27/meta_llama_author_lawsuit/
[21] https://www.ballardspahr.com/insights/alerts-and-articles/2025/07/from-input-to-impact-the-market-harm-standard-emerging-in-ai-fair-use
[22] https://www.theregister.com/2025/02/12/thomson_reuters_wins_ai_copyright/
[23] https://www.reuters.com/legal/litigation/anthropic-reaches-deal-ai-guardrails-lawsuit-over-music-lyrics-2025-01-03/
[24] https://www.theregister.com/2024/05/23/openai_news_corp/
[25] https://www.theregister.com/2024/06/28/openai_google_ai/
[26] https://committees.parliament.uk/writtenevidence/126981/pdf/
[27] https://trumpwhitehouse.archives.gov/wp-content/uploads/2021/01/IPEC-Annual-Intellectual-Property-Report-January-2021.pdf
[28] https://www.opensecrets.org/news/2024/06/lobbying-on-ai-reaches-new-heights-in-2024/
[29] https://democrats-cha.house.gov/media/press-releases/morelles-statement-abrupt-firing-shira-perlmutter-register-copyrights
[30] https://whitepapers.theregister.com/
In [1]a letter [PDF] addressed to US Attorney General Pam Bondi and Michael Kratsios, director of the White House Office of Science and Technology Policy, more than a dozen leaders of conservative organizations urge the government to reject [2]calls to change copyright law to accommodate AI companies.
The signatories argue that supporting the tech industry's effort to reform copyright law would harm American workers, contradict the Trump administration's AI and trade policy, dilute US soft power, and encourage economic espionage by China.
[3]
"Opponents of applying copyright law to AI make a series of flawed claims in an effort to cut corners," the letter says. "One is that competition with China requires the US to abandon its commitment to strong IP protections."
[4]
[5]
The authors contend that weakening IP rights domestically will allow China and other adversaries "to avail themselves of the same dubious 'fair use' theories to not only steal creative content, but also proprietary US AI models and algorithms."
Prior to the debut of OpenAI's ChatGPT in November 2022, legal scholars voiced support for the tech industry's view that training AI models on copyrighted content qualified for the fair use defense against infringement claims.
[6]
The [7]fair use defense under US law requires an analysis of four factors: the purpose and character of the use, the nature of the copyrighted work, the amount of material used, and the effect upon the market for the copyrighted work.
In 2021, for example, judicial law clerk Jenny Quang, who at the time was a UC Berkeley Law School research assistant, wrote an [8]article [PDF] for the Berkeley Technology Law Journal titled, "Does Training AI Violate Copyright Law?"
Quang wrote, "Training a machine learning model with this copyrighted data does not infringe because the data are not redistributed or recommunicated to the public. Copyright protects creative expression, but model training extracts unprotectable ideas and patterns from data."
[9]
That same year, Nat Friedman, then CEO of GitHub, echoed that line of reasoning as GitHub debuted a beta version of its Copilot AI assistant. He [10]wrote in a social media post , "training ML systems on public data is fair use [and] the output belongs to the operator, just like with a compiler."
[11]Amazon is forging a walled garden for enterprise AI
[12]Zig quits GitHub, says Microsoft's AI obsession has ruined the service
[13]China using AI as 'precision instrument' of censorship and repression, at home and abroad
[14]Apple swaps one ex-Google AI chief for another
Various legal scholars have made [15]similar [16]arguments .
But the AI-related copyright claims heard to date by US courts suggest AI companies can't count on a fair use defense in all cases – and there are [17]more than 50 of them presently.
When legal scholar Pamela Samuelson asked the same question in October in [18]an article titled "Does Using In-Copyright Works as Training Data Infringe?", she came to a less definitive conclusion based on the outcomes of [19]Bartz v. Anthropic and [20]Kadrey v. Meta .
In both cases, authors sued the AI providers (Anthropic and Meta) for training on copyrighted books, and while the judges mostly found for the AI companies, the results weren't completely cut and dried. In Anthropic, the judge ruled that as long as the company acquired the copyrighted material legally, it could train its AI on it; Anthropic later settled with authors of some books that it had pirated. In the Meta case, the judge said that it was fair to use the copyrighted texts and the plaintiffs did not present enough evidence of market harm, which at least paves the way for future rulings to take market effects into consideration.
"The Bartz and Kadrey decisions are mixed bags," she wrote. "They indicate that some training data uses may be found fair use, while others may not."
One of the things that has changed in the past few years is that the [21]market impact of AI on copyrighted works is starting to show up.
In Thomson Reuters' copyright claim against Ross Intelligence, the court in February found that training AI models on Westlaw content to build a rival legal service [22]did not qualify as fair use .
AI companies have recognized the risk of infringement claims by [23]settling or striking content [24]licensing [25]deals , though they'd prefer not to incur that cost.
And the potential cost is high because so much of what people have written is subject to copyright protection. As OpenAI CEO Sam Altman said in a December 2023 [26]submission [PDF] to an inquiry by the UK House of Lords, leading AI models could not be trained without the use of copyrighted content.
"Limiting training data to public domain books and drawings created more than a century ago might yield an interesting experiment, but would not provide AI systems that meet the needs of today's citizens," he said.
If AI companies were broadly required to pay for the data they use, their already extensive data center expenditures would balloon with the cost of licensing arrangements.
The conservative letter signatories say that's just fine.
"It is absurd to suggest that licensing copyrighted content is a financial hindrance to a $20 trillion industry spending hundreds of billions of dollars per year," the letter says. "AI companies enjoy virtually unlimited access to financing. In a free market, businesses pay for the inputs they need. Imagine if AI CEOs claimed they needed free access to semiconductors, energy, researchers, and developers to build their products. They would be laughed out of their boardrooms."
Adam Eisgrau, senior director of AI, Creativity and Copyright Policy at Chamber of Progress – one of the tech-aligned groups cited by the conservative authors – rejects the call to have the feds remain on the sidelines.
In an email to The Register , he said, "The purpose of copyright as defined in the Constitution is to promote innovation: what the Framers called the 'progress of science and useful arts.' Under the law, by definition fair use isn't theft or copyright infringement; it's exactly what the Framers wanted. The DOJ absolutely should go to bat for innovators and fair use."
The letter concludes, "President Trump – himself a bestselling author and former television producer – said it best: 'The pioneering spirit of [our] artists, authors, inventors, and other creators has improved our lives and the lives of millions of people around the world, and will continue to propel us toward a better future.'"
Trump made those remarks [27]in April 2020 [PDF], before [28]AI lobbying surged . We note that earlier this year, the President fired Shira Perlmutter, Register of Copyrights, an event that US Representative Joe Morelle (NY-25) [29]linked to Perlmutter's refusal "to rubber-stamp Elon Musk's efforts to mine troves of copyrighted works to train AI models." ®
Get our [30]Tech Resources
[1] https://cdn.prod.website-files.com/65a5dbf7ffda034a47a316a2/692de157f04ef0ddcecc1d34_Coalition%20Letter%20to%20DoJ%20and%20OSTP%20re%20AI%20and%20Copyright_12.1.25.pdf
[2] https://progresschamber.org/resources/white-house-request-for-information-regulatory-reform-on-artificial-intelligencefederal-register-docket-no-ostp-2025-0005/
[3] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2aTDBBznNocGx8l5NdhdgfQAAAMA&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[4] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aTDBBznNocGx8l5NdhdgfQAAAMA&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[5] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33aTDBBznNocGx8l5NdhdgfQAAAMA&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[6] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aTDBBznNocGx8l5NdhdgfQAAAMA&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[7] https://www.copyright.gov/fair-use/
[8] https://btlj.org/wp-content/uploads/2023/02/0003-36-4Quang.pdf
[9] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33aTDBBznNocGx8l5NdhdgfQAAAMA&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[10] https://twitter.com/natfriedman/status/1409914420579344385
[11] https://www.theregister.com/2025/12/03/amazon_enterprise_ai_walled_garden/
[12] https://www.theregister.com/2025/12/02/zig_quits_github_microsoft_ai_obsession/
[13] https://www.theregister.com/2025/12/03/aspi_china_ai_report/
[14] https://www.theregister.com/2025/12/02/apple_ai_chief/
[15] https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4986763
[16] https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5395242
[17] https://chatgptiseatingtheworld.com/2025/10/08/status-of-all-51-copyright-lawsuits-v-ai-oct-8-2025-no-more-decisions-on-fair-use-in-2025/
[18] https://cacm.acm.org/opinion/does-using-in-copyright-works-as-training-data-infringe/
[19] https://www.theregister.com/2025/09/08/anthropic_settles_author_lawsuit/
[20] https://www.theregister.com/2025/06/27/meta_llama_author_lawsuit/
[21] https://www.ballardspahr.com/insights/alerts-and-articles/2025/07/from-input-to-impact-the-market-harm-standard-emerging-in-ai-fair-use
[22] https://www.theregister.com/2025/02/12/thomson_reuters_wins_ai_copyright/
[23] https://www.reuters.com/legal/litigation/anthropic-reaches-deal-ai-guardrails-lawsuit-over-music-lyrics-2025-01-03/
[24] https://www.theregister.com/2024/05/23/openai_news_corp/
[25] https://www.theregister.com/2024/06/28/openai_google_ai/
[26] https://committees.parliament.uk/writtenevidence/126981/pdf/
[27] https://trumpwhitehouse.archives.gov/wp-content/uploads/2021/01/IPEC-Annual-Intellectual-Property-Report-January-2021.pdf
[28] https://www.opensecrets.org/news/2024/06/lobbying-on-ai-reaches-new-heights-in-2024/
[29] https://democrats-cha.house.gov/media/press-releases/morelles-statement-abrupt-firing-shira-perlmutter-register-copyrights
[30] https://whitepapers.theregister.com/