The possibility has always existed that governments—faced with the fact that copyright law does not excuse the mass copying of copyrighted data to train generative AI models—might change the laws to suit AI companies. It’s been a risk even if the law is an obligation of the 1896 Berne Convention, one of our longest-lived international treaties.
I can’t remember where I first heard the “magical pixie dust” analogy for this option — it might have been from Ed Satow, legal scholar and former Human Rights Commissioner in Australia, who has been remarkably clear-eyed about all this.
But I certainly ran with it in my last newsletter, when I saw the thrust of the OpenAI and Google submissions to the Trump administration's request for input in a new AI executive order.
But if we were speaking of sprinking pixie dust before, it is clear that we might instead get truckloads of the stuff, dumped on the law from on high. Evidence in point:
On May 9, the US Copyright Office issued its long anticipated third document in a planned three-document series of papers on generative AI and copyright. This one made the perfectly common sense observation that:
Various uses of copyrighted works in AI training are likely to be transformative. The extent to which they are fair, however, will depend on what works were used, from what source, for what purpose, and with what controls on the outputs—all of which can affect the market. When a model is deployed for purposes such as analysis or research—the types of uses that are critical to international competitiveness—the outputs are unlikely to substitute for expressive works used in training. But making commercial use of vast troves of copyrighted works to produce expressive content that competes with them in existing markets, especially where this is accomplished through illegal access, goes beyond established fair use boundaries.
Some 48 hours later, Shira Perlmutter was fired. She’s the head of the US Copyright Office which issued this document. In this case there was no DEI justification, no screed about unAmerican values, just a two sentences email with no reason given. This all according to Politico.
Let‘s review the timeline:
On Thursday, May 9th, the Copyright Office posts a prepublication version of Part 3 of their mandated AI and Copyright report “in response to congressional inquiries and expressions of interest from stakeholders. A final version of Part 3 will be published in the near future, without any substantive changes expected in the analysis or conclusions.” This includes the statements on fair use from above.
On the evening of Thursday, May 9th, Librarian of Congress Carla Hayden was sacked, in a terse two sentence email from the Presidential Personnel Office. The Copyright Office reports to the Librarian of Congress.
On Friday, May 10th, White House Press Sec offers a statement at the daily press briefing, that “There were quite concerning things that she had done at the Library of Congress in the pursuit of DEI and putting inappropriate books in the library for children.” (Er, all books published in the US are deposited in the LOC… under law…)
On the afternoon of Saturday, May 11th, Perlmutter gets her two-sentence email giving her the sack.
More context:
On Friday, April 11, Jack Dorsey, founder of Twitter called for the abolition of copyright law, well in fact, all intellectual property law - “Delete all IP laws”. Elon Musk replied within minutes “I agree”. 4.9m views on the original Dorsey tweet, and 22,000 likes.
I know some of you receiving this newsletter will be closer to the thick of the action. If you have any news to share, please let me know.