The real threat of AI models training on open-access material is that they may bleed free repositories dry by draining resources and not providing attribution
“Wait, not like that”: Free and open access in the age of generative AI https://www.citationneeded.news/ ... Chosa / @lowficarrots@tenforward.social : https://www.citationneeded.news/ ... super smart essay on what open access means in the age of everything getting scrapped Miguel Afonso Caetano / @remixtures@tldr.nettime.org : “Anyone at an AI company who stops to think for half a second should be able to recognize they have a vampiric relationship with the commons. While they rely on these repositories for their sustenance, their adversarial and disrespectful relationships with creators reduce the incentives for anyone to make their work publicly available going forward (freely licensed or otherwise). … Molly White / @molly0xfff@hachyderm.io : Newsletter: The real threat isn't AI using open knowledge — it's AI companies killing the projects that make knowledge free. The future of free and open access isn't saying “wait, not like that” — it's saying “yes, like that, but under fair terms”. — https://www.citationneeded.news/ ... … Threads: Parker Thompson / @parkert : Content people really hate fair use. There's unlikely to be a “solve” for that, which is great for society, but maybe the bad feels can be reduced. These free an open initiatives greatly benefit tech companies, so it would be great to see them push some money into these ecosystem to help them do more awesome stuff. See also Mediagazer
Context & Ripple Effects
The debate over generative-AI training has largely centered on copyright and fair use, including OpenAI’s defense of training practices in its dispute with the Times, where it characterized training as fair use and offered an opt-out. This commentary shifts the focus to the operating health of the open repositories that supply material, not just the rights claims that follow.
It also extends earlier calls for writers and publishers to seek compensation from major AI companies through collective pressure for a share of AI-created value: attribution and resource burdens can affect whether people keep contributing to openly available collections at all.
First-order effects
- Free repositories and their maintainers can absorb increased access, bandwidth, and moderation demands from model-training activity without a corresponding contribution or attribution mechanism.
- Creators whose work is used without attribution have weaker visibility and fewer incentives to continue placing work in open-access channels.
Second-order effects
- Repository operators may respond with tighter access controls, rate limits, licensing terms, or requests for financial support, raising friction for AI developers that rely on broadly available corpora.
- The dispute broadens from whether model training is lawful to who funds and maintains the underlying commons, adding a practical dimension to existing compensation campaigns.
Third-order effects
- If open repositories cannot sustain uncompensated machine use, open access could become less dependable as shared AI input infrastructure and more segmented by access rules or commercial arrangements.
- The longer-run contest is likely to concern governance of the commons—attribution, support, and permitted reuse—rather than copyright litigation alone.
The trend: Generative AI is turning openly accessible cultural and research material into infrastructure whose upkeep, credit, and access rules are increasingly contested.