AI-powered song generator startups Udio and Suno have been hit with copyright infringement lawsuits from the Recording Industry Association of America (RIAA). The lawsuits, filed by major labels including Sony Music Entertainment, Universal Music Group, and Warner Records, claim that the companies trained their generative AI models using copyrighted music without compensating the labels. The RIAA argues that the proliferation of machine-generated content could saturate the market, diluting the value of genuine sound recordings. The litigation adds to a growing body of cases against generative AI vendors, which raises questions about the fair use of copyrighted data in machine learning training. The outcome of these lawsuits remains uncertain and could have a significant impact on the future of AI development.
One possible precedent for the courts to consider is the case of Google Books. Over a decade ago, Google scanned millions of books to create an online archive. Authors and publishers sued Google for copyright infringement, but the courts ultimately ruled that Google Books’ copying had a “highly convincing transformative purpose.” This ruling could potentially be applied to generative AI models if it can be demonstrated that they do not plagiarize copyrighted works on a large scale. Alternatively, judges may assess each model and its generated outputs on a case-by-case basis. The outcome of these lawsuits will be closely watched by AI vendors whose business models rely on the use of copyrighted data.
In other news, OpenAI has delayed the release of its advanced Voice Mode feature for its AI chatbot platform, ChatGPT. However, OpenAI has recently acquired remote collaboration startup Multi and launched a macOS client for all ChatGPT users. Stability AI, the maker of open image-generating model Stable Diffusion, has been saved from financial trouble by a group of investors that includes Napster founder Sean Parker and former Google CEO Eric Schmidt. The company has also appointed a new CEO as part of its efforts to regain stability in the competitive AI landscape. Google is rolling out a new AI-powered side panel in Gmail, called Gemini, which offers writing assistance and summarization features. Otis Chandler, co-founder of Goodreads, has launched Smashing, an AI-powered content recommendation app that helps users discover hidden gems on the internet.
A study by NewsGuard has found that leading AI chatbots, including ChatGPT, Anthropic’s Claude, and Google’s Gemini, regurgitate snippets from Russian propaganda campaigns. The chatbots responded with disinformation 32% of the time when asked about narratives created by Russian propagandists. This study highlights the increased scrutiny on AI vendors as the US election approaches. Leading AI companies have agreed to take action to curb the spread of deepfakes and election-related misinformation, but platform abuse remains a challenge.
Researchers at MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) have developed a model called DenseAV that can learn language by predicting what it sees from what it hears, and vice versa. Trained on a dataset of 2 million YouTube videos, DenseAV can identify objects from their names and sounds. The researchers hope to create systems that can learn from massive amounts of video or audio data in the future.
OpenAI CTO Mira Murati has admitted that generative AI could eliminate some creative jobs. While acknowledging that job losses are possible, she also suggested that some jobs “maybe shouldn’t have been there in the first place.” This statement has drawn criticism from creatives who argue that OpenAI and other AI companies are profiting from the works of artists without proper compensation. OpenAI has promised to release tools to give creators greater control over how their works are used and has signed licensing deals with copyright holders and publishers. However, critics argue that more needs to be done to address the impact of AI on job markets and to support affected workers.
