AI models should be changed to alleviate the risk of piracy

A motion, which I proposed, was passed at the Royal Pharmaceutical Society (RPS)’s Annual General Meeting on 4 June 2025 asking the Society to issue a public statement of its policy in relation to AI and piracy of copyrighted materials.

Several members of the RPS have had books pirated by Meta and an earlier edition of Martindale has also been pirated. AI firm Anthropic has now agreed to pay US$1.5bn (£1.11bn) to settle a class action lawsuit filed in the United States by authors who said the company stole their copyrighted work to train its AI models. 

The RPS has now issued a position statement — ‘Artificial intelligence, copyright and intellectual property’ in December 2025.

I have also suggested that it would also be useful for the RPS to publish advice to members on how to prevent AI companies pirating material intended to be covered by copyright. One key aspect is that AI software, such as generative AI and large language models, all have privacy and data control settings, which by default assume that any material it provides can be used in further development of its model.

Therefore, if a research paper or the text of a book is prepared with the aid of one of these AI software models, the contents of the book or paper will automatically be included in the model unless the privacy and data control toggles in settings are changed to obviate this.

Submission of a research paper intended for a journal, such as The Pharmaceutical Journal or any of the other RPS journals, which will then claim copyright, is pointless unless these changes have been made to the AI model used — in effect, it is bolting the stable door after the horse has bolted.

Tony Cartwright

Last updated
Citation
The Pharmaceutical Journal, PJ January 2026, Vol 316, No 8005;316(8005)::DOI:10.1211/PJ.2026.1.393296

    Please leave a comment 

    You may also be interested in