Context: Yesterday the first major Indian copyright lawsuit against an AI provider surfaced as the Delhi High Court (HC) listed a hearing in ANI Media v. OpenAI (ANI = Asian News International) for today (November 18, 2024 ai fray article).
What’s new: OpenAI’s Indian counsel just declared in open court and for the record that as of October 2024, OpenAI has ceased to use ANI’s content for training. That declaration on OpenAI’s behalf will be recorded in the court order following the hearing. The information was just shared on LinkedIn by Indian lawyer Sneha Jain (a partner at the Saikrishna & Associates firm).
Direct impact: OpenAI’s lawyer stressed that this measure was taken “without prejudice” to their defenses. But it may be a worldwide first. Someone who is confident of not running afoul of a given jurisdiction’s copyright laws would normally stand their ground and not give the impression of being forced to cease and desist from an action previously taken. It suggests that OpenAI is unsure of its legal position in India. It is, however, important to note that Indian copyright law (as explained in the previous ai fray article) differs from, for example, U.S. copyright law.
Wider ramifications: OpenAI’s strategic retreat potentially encourages other Indian right holders to sue, and despite differences between the legal frameworks, it could embolden right holders in other jurisdictions to seek similar concessions.
It must have been a difficult decision for OpenAI. They can certainly argue that there are enough other content sources in India. If one asks ChatGPT for the leading Indian news agencies, ANI is number three on a list of five. But there is nothing stopping the other agencies from asserting their rights in the same way as ANI, and OpenAI’s concession, albeit without prejudice, does nothing to discourage others from bringing their own copyright infringement claims.
It is also striking that OpenAI made this decision so early in the process. Until yesterday, it was not even known that ANI had sued OpenAI.
To the best of ai fray‘s knowledge, no similar concessions have been made in other jurisdictions, neither by OpenAI nor by other AI providers. There is no indication of Anthropic having stopped using certain books as training material only because of a book author class action in the U.S. (August 20, 2024 ai fray article) or of OpenAI having stopped using New York Times content (July 2, 2024 ai fray article).
In terms of the effect OpenAI’s strategic retreat has on the court, there is a formal dimension as well as a psychological one:
- In formal terms, it means that the court is going to consider the question of injunctive relief less urgent (if not moot). Also, it limits the potential damages period and, therefore, the amount of the potential damages claim.
- In psychological terms, it suggests respect for India’s laws, which is a good thing, but it also shows that OpenAI is not 100% certain of the strength of its defenses, a fact that won’t be lost on the court either.
Approximately three dozen AI copyright cases are pending in the United States, a jurisdiction with a fair use exception that is traditionally strong and has been outcome-determinative in some high-profile recent tech copyright cases (Google Books as well as Oracle v. Google). For right holders that primarily generate revenues in the U.S. market, such as the New York Times, the obvious forum choice is to sue in a U.S. district court. Theoretically, damages awards in the U.S. could be high, though outsized jury awards are often tossed or adjusted on appeal.
Now there are more and more international right holders filing suit in their jurisdicitons, with different legal frameworks. The fact that ANI achieved without a fight what no U.S. litigant appears to have accomplished so far is attributable to the fact that copyright enforcement is rather country-specific. The first major European copyright lawsuit against OpenAI is pending in Munich, Germany (November 14, 2024 ai fray article).