The Criticized Microsoft Boss Says Internet Content Is Free To Train AI

RediksiaMonday, 1 July 2024 | 13:07 GMT+0000

Diksia.com - Microsoft AI CEO Mustafa Suleyman made a controversial statement that drew criticism from various groups, especially content creators, saying that the internet essentially serves as “free software” for training AI models.

In an interview session on NBC News, the host expressed his concerns about training AI using online content and mentioned that OpenAI uses recorded YouTube videos to train its models. 

Mustafa was then asked who should own intellectual property (IPR) in such a situation and how the commercial agreements around it should be structured.

“In terms of content that already exists on the open web, the social contract of that content since the ’90s has been that it’s fair use. Anyone can copy it, recreate it, reproduce it. It’s already freeware (software-free), if you will. That’s the understanding,” he replied.

Mustafa’s comments suggest that AI developers are free to use the vast amounts of data available online to train their models. This view seems to ignore the complex legal and ethical issues surrounding ownership and usage rights of content.

Fair use allows limited use of copyrighted material for purposes such as teaching or research. However, the use of large amounts of content to develop AI models goes far beyond these limits, especially when there are clear commercial motives at play.

Artist Denman Rooke highlighted the difference between viewing or downloading art online and using it commercially without permission, stressing that this is theft.

FYI, one of the biggest controversies surrounding AI chatbots like ChatGPT, Gemini, and Copilot is that generative AI companies may be scraping proprietary data and using it to train their AI models.

In recent months, several organizations and publications such as Forbes, the New York Times and the Recording Industry Association of America have filed lawsuits against parties such as Microsoft, OpenAI, Perplexity, Udio and others, alleging that these companies used content from them to train their AI models without permission, Indianexpress reported on Monday (01.07.2024).