Microsoft Execs Admit to Creating AI-Driven Doom Loop in News Industry

The New York Times has been suing OpenAI and Microsoft for nearly three years over allegations that their large language models, like ChatGPT, have infringed on the newspaper's copyright.

The lawsuit claims that these AI models have trained extensively on millions of New York Times articles from decades past without obtaining permission or paying licensing fees.

In a recently unsealed legal filing, an unnamed Microsoft executive admitted to creating a 'doom loop' that threatens both the journalism and tech industries. The executive stated that their AI content strategy has started a cycle where the performance of these models will be hurt by the economic foundations of their essential suppliers.

The Times' attorneys have also quoted defendants' own numbers, showing that Bing click-through rates on New York Times articles have dropped 83-93% since the widespread adoption of its AI news summaries. This has led to a self-defeating logic, where LLMs (large language models) are seen as products that destroy their supply chain.