The balance between software copyright protection and openness—achieved through 40 years of debate—has been disrupted by generative AI, according to one analysis. The concern centers on how large language models consume online content without regard for existing licenses and copyleft agreements.
The modern internet infrastructure, including cloud services and the backbone of online connectivity, was built on free and open-source software that allowed derivative works to flourish. This openness created an equilibrium: developers could learn from shared code, contribute improvements, and pass those rights forward through licenses like GPL, MPL, and Creative Commons SA. Without this model, the analysis suggests, the internet would resemble AOL or CompuServe—far more expensive and concentrated in the hands of a few companies.
However, generative AI has disrupted this arrangement. According to the source, LLMs are “consuming everything they find online, without regard for copyright or license,” and “there seems to be no appetite for legal enforcement of these obligations.” This represents what the author calls a broken “social contract.”
The consequences extend beyond copyright violations. Developers now face competing incentives: sharing code openly enables AI systems to more easily discover vulnerabilities in their work, while keeping code closed makes those mistakes harder to find. Publishing code on platforms like GitHub creates another problem—the author notes they may “be inundated with pull requests generated by AI bots and inexperienced users, flooding me with mostly useless slop.”
There are also concerns about trustworthiness. Code shared online may have been “polluted by slop code,” poisoned with malicious libraries, or comprised of stolen work from others—implicating users in potential copyright violations.
The broader worry is that AI consolidates wealth and power from those who created the underlying work. The analysis argues that “the act of openness will be twisted and abused into creating more consolidated wealth and power, without respect or credit for those who did the real labour.” This threatens the foundation of knowledge-sharing that, according to the source, has transformed the world “in a mostly positive and empowering way” over the past 35 years.
Key facts
- Large language models are consuming online content without regard for copyright or license restrictions
- The modern internet infrastructure was built on free and open-source software enabled by copyleft licenses like GPL and MPL
- Developers sharing code publicly now risk enabling AI systems to discover vulnerabilities in their work
- Existing copyright and license agreements have been enforceable in courts, but there appears to be no appetite for legal enforcement against AI training practices
- Open-source sharing has been foundational to internet development over the past 35 years