Major technology corporations are currently generating massive profits by utilizing the very internet standards and open-source frameworks developed by Aaron Swartz. While Swartz faced aggressive federal prosecution for downloading academic articles, companies like Meta and Anthropic now face legal scrutiny for allegedly harvesting millions of copyrighted works to train proprietary large language models.
The tension between the democratization of knowledge and the commercialization of data has reached a breaking point. The industry is grappling with the fallout of a massive copyright settlement and lawsuits that highlight a stark double standard. The infrastructure of the modern AI era is built upon a foundation of free software and decentralized standards that Swartz helped pioneer, yet the economic benefits of these technologies are being concentrated within a handful of trillion-dollar industry players.
Markdown and Python underpin the AI stack
Swartz’s technical fingerprints are embedded in the core workflows of generative AI. In 2004, Swartz collaborated with John Gruber to introduce Markdown, a lightweight text-formatting syntax. For instance, OpenAI recommends using Markdown to organize instructions and provide structural clarity to ChatGPT models. This format allows users to communicate complex hierarchies to LLMs without the overhead of proprietary markup languages.
Beyond text formatting, the execution layer of the AI revolution relies on the Python ecosystem. Swartz developed web.py, a public domain framework for building Python applications. Today, Python is the lingua franca of machine learning. Systems such as Microsoft’s GitHub Copilot, Claude Code, and ChatGPT utilize Python to execute code, automate complex data analysis, and manage the heavy lifting of software generation. The transition from Swartz's open-source contributions to the proprietary dominance of AI agents changes how technical agency is distributed.
Reddit data monetization creates a value gap
The commercialization of human interaction has turned community-driven platforms into gold mines for model training. Swartz’s startup, Infogami, merged with Reddit in 2005, making him co-owner of what would become a massive repository of human thought. For years, users contributed to these digital commons with little expectation of being commodified. That changed in 2024, when Reddit entered into a reported $60 million annual agreement with Google to provide access to its pages for machine learning purposes.
The same year, Reddit formalized a partnership with OpenAI, granting the company access to Reddit’s data to incorporate into ChatGPT. This move effectively converts the collective experiences and opinions of millions of users into proprietary intellectual property.
Legal disparities expose corporate data harvesting
The contrast between the treatment of Swartz and the current legal battles involving Big Tech is mathematically staggering. In 2011, federal prosecutors charged Swartz with computer fraud after he used MIT’s network to download millions of academic articles from JSTOR. Swartz argued that scholarly knowledge should be public, yet he faced the prospect of decades in prison before his death in January 2013. While he was not charged with copyright infringement, the weight of the federal prosecution was devastating.
In contrast, the scale of data acquisition by modern AI labs is unprecedented. Recent legal filings and settlements reveal a different reality for corporations:
| Entity | Alleged Action / Settlement | Context |
|---|---|---|
| Anthropic | $1.5 billion copyright settlement | Acquired over seven million unauthorized copies from LibGen and Pirate Library Mirror. |
| Meta | Lawsuit by Elsevier (May 2026) | Alleged download of millions of copyrighted books and scholarly publications, including material originating from Sci-Hub. |
| Meta | BitTorrent Allegations | Accused of using BitTorrent to download over 81.7 terabytes of “pirated” academic papers. |
Where Swartz’s actions were viewed as a criminal threat to intellectual property, these corporate activities are often framed as high-stakes litigation between behemoths. The sheer volume of data—terabytes of academic research and millions of books—makes the 2011 JSTOR incident appear as a drop in the ocean compared to the massive, automated ingestion of the written word by Llama and other proprietary models.
The erosion of decentralized standards
Swartz was a staunch advocate for internet standards that prevented central gatekeepers from controlling information flow. His 2002 Standards Manifesto emphasized the importance of protocols like RSS 1.0, which allowed for the decentralized distribution of news and media. This architecture gave rise to the podcasting industry, enabling creators to bypass traditional media intermediaries through open feeds.
However, the industry has trended toward “walled gardens.” Apple, Spotify, and Google’s YouTube have become dominant forces in this multibillion-dollar industry, often marginalizing the very decentralized protocols that allowed them to grow. This centralization mirrors a trend in the AI era: the move from open, interoperable standards to closed, proprietary ecosystems. Even as companies utilize open-source tools like Python and Markdown, they increasingly restrict the user’s ability to interact with the resulting technology on their own terms.
Data collection for LLMs blurs privacy and surveillance lines
The drive to capture data for LLMs intersects dangerously with existing privacy concerns. The FTC has already identified extensive surveillance practices by large social media and streaming companies. When these same companies act as the primary collectors of training data, the line between “improving a model” and “surveilling a user for profit” becomes non-existent. The knowledge of millions is being ingested into systems that are simultaneously used to monitor and profile those very same individuals.
The fundamental question remains: who owns the digital commons? As we move further into an era defined by generative intelligence, the tension between the free software principles Swartz championed and the proprietary hunger of the AI industry continues to escalate. Whether the legal system will eventually enforce a “right to participate” in the value created by collective human knowledge, or simply facilitate its corporate enclosure, remains the defining unknown of the decade.
