Internal documents reveal AI industry concerns over economic self-sabotage

Recently unredacted legal documents and depositions in the New York Times lawsuit against Microsoft and OpenAI have disclosed that senior executives were aware of the potentially destructive economic impact of artificial intelligence. The internal records indicate that leaders understood the risks associated with rapidly launching products featuring the technology following the widespread success of ChatGPT. These documents describe a phenomenon termed a doom loop, a situation in which AI systems monetise the work of human creatives while simultaneously undermining the economic structures that originally funded that creative output.

One internal Microsoft document quoted by the newspaper’s legal team warned that the company’s AI content strategy had initiated a cycle that would harm both the performance of its models and the broader web. The note described it as highly unusual for an end product to threaten the economic foundations of its essential suppliers, yet stated that this was the situation created for the large language model business regarding its content supply chain. This internal acknowledgement highlights a fundamental contradiction within the industry, where the products being pushed to users rely on fresh cultural content even as they erode the economic incentives required to create it.

The implications of this dynamic are becoming evident across various sectors. In journalism, the traditional model of publishing work under the assumption that online platforms would drive readership is under strain. Tools such as ChatGPT and Google’s Gemini now summarise articles without directing users to the original sites, sparking a crisis that threatens the viability of independent media. Similar pressures are being felt in music and filmmaking, where AI-generated content is increasingly crowding out human-produced work. In the music industry, streaming services are grappling with how to manage the influx of AI-generated tracks, with some implementing strict rules while others continue to allow users to exploit algorithms. Hollywood studios are also increasing their reliance on AI, a move that has prompted protests from actors who fear their likenesses and voices could be rendered worthless, posing an existential threat to their livelihoods.

The job market is experiencing its own version of this cycle. Desperate job seekers are increasingly using AI to enhance their resumes and navigate AI-powered recruitment portals, while recruiters face an overwhelming volume of nearly identical applications. Daniel Chait, chief executive of applicant tracking system company Greenhouse, described this as a tragic situation where both sides are using AI to solve their problems in ways that make the issue worse. He noted that increased use of the technology begets more use, to no one’s benefit, creating a scenario where the more it happens, the worse it gets.

Education is also undergoing significant disruption. Students are using AI to complete homework, offloading cognitive tasks to machines, while teachers are using the technology to generate assignments. This shift has placed educators who oppose the use of such tools at odds with school administrators. A recent study by the Brookings Institution’s Center for Universal Education concluded that the risks of using generative AI in education massively outweigh the benefits. Educators have warned that college students are rapidly losing the ability to read, suggesting that the impact on learning outcomes is severe.

Proponents of AI, including the legal teams for OpenAI and Microsoft in the ongoing New York Times lawsuit, argue that the technology brings new value to the table. They contend that the output of AI models is sufficiently differentiated from the original content it draws upon to fall under the doctrine of fair use in US copyright law. However, the validity of this argument in court remains uncertain. Precedents, such as the settlement by Anthropic for billions of dollars after being found to have violated copyright laws, suggest that legal risks are significant. With more than 200,000 jobs in creative industries lost over the last four years, according to the US Bureau of Labor Statistics, the economic fallout is substantial. The AI industry appears to be trapped in a cycle of its own creation, undermining the very economic systems that produced the data it needs to exist, a strategy that may not be sustainable in the long run.