OpenAI searches for an answer to its copyright problems

🌈 Abstract

The article discusses the deals made between OpenAI and major publishers like Axel Springer, Condé Nast, and The Associated Press, as well as the legal implications and potential impact on the future of web search.

🙋 Q&A

[01] The Deals

1. Questions related to the content of the section?

• Why would OpenAI pay for access to content it had already scraped from the web without permission?
• Why would publishers, some of whom are angry about their work being used without permission, agree to these deals?
• What are the key terms of the deals between OpenAI and the publishers?
• How much money is OpenAI paying the publishers, and what is the ceiling on these deals?
• What is the purpose of these deals from OpenAI's perspective?

Answers:

• OpenAI may be paying the publishers to ensure they don't sue over the content it had already scraped without permission. The deals can be seen as settlements to prevent lawsuits.
• The deals give the publishers some leverage and may eventually force Google to the negotiating table, as Google has also been ingesting publisher content without permission.
• The details of the deals have remained murky, likely due to non-disclosure agreements. Publicly reported figures suggest OpenAI is paying as little as $1-5 million per year per publication, with a ceiling of around $10 million per year.
• From OpenAI's perspective, the deals provide up-to-date information to improve the capabilities of its products like ChatGPT and SearchGPT, as well as help manage its reputation and legal risks.

[02] The New York Times Lawsuit

1. Questions related to the content of the section?

• What is the key allegation in the New York Times' lawsuit against OpenAI?
• What are the potential damages the Times could be awarded if it wins the lawsuit?
• How do the publisher deals relate to OpenAI's fair use defense in the lawsuit?

Answers:

• The New York Times alleges that OpenAI engaged in copyright infringement by ingesting the Times' work to train its large language models, and that the resulting products now compete with the Times.
• If the Times wins, it could be entitled to statutory damages of at least $7.5 billion, as well as compensatory damages, restitution, and attorneys' fees.
• The publisher deals can be seen as a way for OpenAI to try to strengthen its fair use defense by securing agreements with the publishers, even though the deals may undermine that argument.

[03] Implications and Outlook

1. Questions related to the content of the section?

• How might a ruling in favor of the New York Times impact the broader AI ecosystem?
• What are the concerns about the potential anticompetitive effects of the publisher deals and copyright lawsuits?
• What are the open questions about the long-term viability and sustainability of AI-powered "answer engines" like SearchGPT?

Answers:

• A ruling against OpenAI could make it prohibitively expensive for smaller players to compete in the large language model space, further entrenching the dominance of tech giants like Google, Microsoft, Amazon, and Meta.
• There are concerns that the publisher deals and copyright lawsuits could narrow the competitive dynamics in the AI ecosystem, as only the largest companies may be able to afford the necessary licensing and legal costs.
• It's unclear whether people actually want "answer engines" instead of traditional search, and whether these AI-powered services can be financially sustainable in the long run.