- The New York Times sued OpenAI and Microsoft for copyright infringement over the use of articles to train AI chatbot ChatGPT.
- The case centers on fair use provisions – AI companies argue models learn patterns to produce new text rather than copying content.
- Previous media versus tech battles were resolved via licensing deals and concessions rather than definitive legal judgements.
- The core dispute is over business impacts and cash flows rather than AI disruption of content models.
- A compromise allowing AI training with licensing payments to publishers is the most likely outcome.
The New York Times recently filed a lawsuit against OpenAI and Microsoft for copyright infringement, claiming the companies’ use of millions of articles to train AI models threatens the Times’ business model. This complaint comes after similar ones from creators such as comedian Sarah Silverman regarding the unauthorized use of copyrighted content.
The case centers on fair use provisions in copyright law, which allow for exceptions for educational purposes and transformative works but offer stronger protections for creative works. Impact will likely be a key factor as the court weighs the competing interests. While OpenAI’s ChatGPT now boasts over 100 million users compared to the Times’ 10 million subscribers, there are counterarguments favoring the AI companies as well.
Generative AI
Generative AI models like ChatGPT do not copy content verbatim from training data but rather learn linguistic patterns, styles, and structures to produce new output. The models aim to gain a conceptual understanding of text rather than reproduce protected expression. As such, OpenAI and Microsoft can claim their systems do not infringe on copyrights but rather develop skills through exposure to vast datasets.
However, media companies can counter that argument by pointing to the sheer scale of scraped content as well as ChatGPT’s ability to mimic stylistic quirks of authors, threatening markets for creative work. The Times may also highlight the potential long-term impacts on its subscription revenues if readers come to view AI chatbots as substitutes for news and analysis.
Previous battles between media and tech groups suggest a compromise allowing AI companies to use copyrighted content for training purposes while providing licensing fees or other concessions to publishers is the most probable outcome. For example, OpenAI already has content licensing agreements with publishers like Axel Springer and The Associated Press to access articles for model development.
Some media outlets are reportedly negotiating similar deals with tech giants like Apple as well. This precedent indicates cash flow and business impacts are likely the core focus rather than definitive legal judgements on AI capabilities versus copyright law.
Viacom’s failed 2007 lawsuit against YouTube for over $1 billion in damages shows the complexities of copyright disputes involving disruptive technologies. While Viacom argued YouTube violated copyright on a massive scale by hosting pirated content, the court ruled that the Digital Millennium Copyright Act safe harbor provisions protected YouTube from liability over users uploading infringing material.
That ruling established firmer protections for platforms that host third-party content. However, after losing, Viacom still secured concessions in a settlement deal. For example, copyright owners could request expedited removal of infringing clips.
A similar outcome leaving room for AI development while providing remedies for publishers is possible in the present dispute. Impact matters more than technicalities around copying versus learning patterns. Generative AI has been trained on internet-scale datasets that do not neatly map to existing copyright frameworks.
In essence, a court showdown would open up complex questions around measuring the value generated by AI systems versus the content they ingest. But rather than a definitive judicial reckoning, the Times’ lawsuit is unlikely to progress that far. As with Viacom and YouTube, a compromise deal remains the most likely finale. The rise of ChatGPT itself shows rapid pace of progress in AI, underscoring why expensive, prolonged lawsuits often give way to licensing agreements and carve outs allowing mutually beneficial coexistence.
Two products launched right now:
AI Tool Agent: https://orbitmoonalpha.com/shop/ai-tool-agent/
AI Tool Drawsth: https://orbitmoonalpha.com/shop/ai-tool-drawsth/