AI has been besieged by nearly 10,000 writers. How will the copyright war shape the direction of the AI industry?

Source: Financial Associated Press

Author: Malan

The copyright war against artificial intelligence has ushered in a new phase, with 8,500 writers in the United States asking artificial intelligence companies to pay for their copyright losses in a joint letter on Tuesday.

The co-authors include Pulitzer Prize-winning novelists Jennifer Egan, Michael Chabon and Louise Erdrich, among others, and the addition of these well-known authors also adds weight to the copyright war.

The joint letter wrote that it is unreasonable that millions of copyrighted books, articles, poems and other works have become the nourishment of artificial intelligence without paying any fees. AI received billions of dollars in development funding and should now be compensated for the use of works.

The letter is not aimed at a single company, and OpenAI, Microsoft, Meta and other AI companies are all within the scope of the warning. The American Writers Guild stated that without compensation, writers will be unable to continue to create, and artificial intelligence can only learn from existing stories and become more and more mediocre.

Persistent copyright issues

Although artificial intelligence has greatly promoted the attention of this technology in the market, over time, more and more people have begun to realize that the copyright issues behind artificial intelligence may be quite difficult.

The Writers Guild of America requires artificial intelligence companies to do three things: obtain the author's permission before using the material; fairly compensate the loss of the author who has used the material in the past and now; compensate the loss caused by the artificial intelligence output related copyright content.

At present, the association has not made any legal threats, and the association's CEO Mary Rasenberger pointed out that due to the huge costs of litigation, writers need a long time to prepare for litigation.

On the other hand, artificial intelligence companies are cautiously silent on copyright issues, and no company wants to be the first bird. Even the U.S. government is in a dilemma about this contradiction.

Last week, the U.S. Congress discussed the relationship between artificial intelligence and copyright protection. Ben Brooks, director of public policy at Stability AI, who participated in the hearing, revealed that the company received more than 160 million withdrawal requests because creators did not want their images to be censored. AI models are used for training.

Asked whether he should pay, Brooks evaded saying that developing models requires data diversity.

Senator Marsha Blackburn slammed the so-called fair use of data as an effective way to steal intellectual property.

possibility

The international discussion on this issue is also quite tortuous. As governments of all countries want to develop their own artificial intelligence industries, they all have a mentality of avoiding data copyright issues. So far, only the United Kingdom has indicated that it will relax the rules on the use of copyrighted materials.

However, although the United Kingdom allows AI companies to use materials to train artificial intelligence models without the permission of right holders, there are blurred boundaries for legally accessed data, which means that the conflict between artificial intelligence and copyright is still prominent.

One draws on music's copyright wars in the early 2000s, when the Napster player gained popularity with nearly everyone, but its ignorance of copyright sparked the ire of the industry's major record labels. In the end, all the stakeholders sit down and negotiate, with the company stepping in to negotiate the licensing deal and define how to legally import the content.

This has also greatly changed the rules of the game in the music industry. There is almost no place for free players to survive. The current music software pays copyright fees to copyright owners through fees.

In contrast, the current artificial intelligence industry, such as ChatGPT, has maintained operations through fees, but the controversial point is that it does not take into account the cost of copyright use in the fees.

Ryan Khurana of AI start-up Wombo believes that artificial intelligence is likely to evolve a licensing system similar to music. But lawyer Matthew Butterick sees similar ideas as disastrous, given the broader scope of artificial intelligence.

Others have suggested that perhaps a fund could be set up to compensate those affected by the training of AI models. But another problem arises: artists are not willing to pay for the one-time license fee, and there is a huge gap between the income of sharing and buyout.

More acutely, technologist Andy Baio points out, artists are constrained by expensive litigation, which could turn copyright wars into protracted wars of attrition. Therefore, the results of the first few copyright lawsuits of artificial intelligence copyright will be crucial and will become the wind vane of this issue.