There have long been concerns about the power of generative AI, in terms of the damage it could do to society, individuals and businesses.
On Wednesday, a major media organization claimed that the emerging technology stole its work.
The New York Times filed a federal lawsuit this week against both Microsoft and OpenAI — the company behind the ChatGPT platform — alleging copyright infringement.
According to the lawsuit, the two companies trained automated chatbots by having them read millions of articles produced by the Times. Now, the technology has begun generating content on its own that is competing with the content the Times creates.
The Times is seeking “billions of dollars in statutory and actual damages” over the “unlawful copying and use of The Times’s’ uniquely valuable works,” the lawsuit states.
All AI companies use information that’s readily available online to train their chatbots, which are powered by generative AI technology. This, of course, includes articles that news organizations publish.
While this material helps to improve the performance of the chatbots, it also can result in the chatbots stealing information and re-purposing it, and even getting some information incorrect.
One example, according to the lawsuit, is that GPT-4 attributed some product recommendations incorrectly to Wirecutter, which the news organization claims endangers its reputation.
.@nytimes sues @OpenAI and @Microsoft for copyright infringement
— This is a big deal because it could set a precedent for 1. How courts define the value of news content in training large language models and 2. What the damages are for previous use@axioshttps://t.co/RPvSYunK4A
— Sara Fischer (@sarafischer) December 27, 2023
While The Times lawsuit isn’t the first filed against AI companies, it is the largest to date.
In addition, a group of 4,000 writers penned a letter to the leaders of Meta, Microsoft, Google, OpenAI and other AI companies to accuse them of using exploitative practices as part of their plan to build their chatbots to “mimic and regurgitate” the ideas, style and language of the writers.
Yet, some news organizations have taken a slightly different approach to generative AI.
The Associated Press, for instance, announced an agreement with OpenAI in June that allows the company to license the full news story archive the AP has produced. This gave the AP a way to profit off the repurposing of its past work, rather than trying to fight to block it in court.
The outcome of the lawsuit could have wide-ranging implications not just for media organizations but for almost any sector potentially affected by generative AI.