Tracey Armstrong, in an interview for The Scholarly Kitchen last week:
What keeps me up at night is the unlicensed use of copyrighted material in AI systems and the lack of recognition globally of the critical, foundational, and perpetual role that copyrighted material plays in powering AI systems. I am focused on ensuring that we have viable market-based licensing solutions for the use of copyrighted content in the training of AI systems and on elevating awareness of the critical importance of using quality content in training, and of the fact that copies are being made in that training process.
The publishers that CCC represents aren't kept up at night. No, they're dreaming of licensing revenue from OpenAI and its big-tech peers. It’s almost certainly the case that the marquee LLMs have trained on giant scholarly publishers’ copyrighted content. Last week’s New York Times report on the tech firms’ “cut corners” testifies to their insatiable demand for materials. CCC and its publisher-clients are, no doubt, preparing lawsuits and negotiating—with the aim, either way, to profit yet again off the unpaid scholarship of researchers.
A side note: Armstrong, in The Scholarly Kitchen interview, claims that CCC is a non-profit. That’s misleading at best. In the early 1980s the U.S. courts forced the firm to drop its non-profit status, since it had no “interests of any substance beyond the creation of a device to protect their copyright ownership and collect license fees.” As the Copyright Society notes, “CCC still maintains not-for-profit status in New York, but for federal purposes is a for-profit corporation.”