Earlier this week, Ed Newton-Rex, a composer and technologist, wrote on X that he was leaving his position as vice president of audio at StabilityAI, a Palo Alto-based startup best known for the image-generation software Stable Diffusion.
“I’m proud that we were able to launch a state-of-the-art AI music generation product trained on licensed training data, sharing the revenue from the model with rights-holders,” Newton-Rex wrote. However, he wrote he’d grown increasingly concerned with the company’s “prevailing opinion on fair use at the company.” StabilityAI took the position that using the work of artists, writers, and musicians to train its generative A.I. models should be considered “fair use” under copyright law — and that those creators are not entitled to compensation.
To date, many companies have used copyright information scraped from the internet or specialized image sets to train their models — or built tools off of those models. The recent strikes by Hollywood writers and actors were in part inspired by concerns their work could be replaced by A.I. systems.
In September, a group of authors filed a copyright lawsuit against OpenAI, creator of the popular tool ChatGPT, for using their work to train OpenAI’s systems. If the lawsuit succeeds, it could “fundamentally reshape the industry,” James Grimmelmann, professor of digital and information law at Cornell University Law School, told ABC News.
Regulators, particularly in the U.S. and EU, are beginning to look at how to update copyright law for the age of generative A.I. — as well as whether the tools are sufficiently protecting consumers and vulnerable populations from harm.
Venture capitalist Linus Liang told Inc. earlier this year that generative A.I. has made it cheaper and faster than ever to launch a tech company. But how many of those companies have staying power is an open question, particularly if the tools they use become more expensive or subject to tighter regulation.
So far, funding to artificial intelligence startups has been a bright spot in an otherwise difficult landscape. Overall venture capital funding was down 50 percent in the first half of the year, but a quarter of the money that has been invested this year has gone to A.I. startups, about double the rate of last year, according to Crunchbase. But many see a bubble that could soon result in a correction, especially since so many companies are working on similar products.
“A thinning of the herd is guaranteed,” Des Traynor, co-founder and chief strategy officer of A.I.-powered customer service platform Intercom, told Inc. in July.