Microsoft AI CEO: Your internet stuff is 'freeware' feed for training models

Mustafa Suliman, a Microsoft AI CEO, reported that most of the content published by automatic learning companies on the Internet this week, and that it could be used for neural network training.

This follows the eight newspapers that Openai and Microsoft continued for the illegal budget for the content in April, like the New York Times four months ago. After that, there are two authors who continued Openai and Microsoft in January. Furthermore, in 2022, based on a claim that some unknown promoters used a public programming code for the training of the generated model in violation of license conditions, to Openai and GitHub. Bred legal measures.

Interview with Andrew Ross Sorkin from CNBC to ASPENE IDEAS Festival, the company is actually engaged in the AI ​​field.

 "In connection with the content on the open web, he thought that the content of this content since the 1990s was fair. “Anyone can copy it, recreate it, reproduce it, if they want, it’s free software. It was understanding.”

Suleiman admitted that there was another category of content: documents published by companies with lawyers.

"There is a separate category where a website, publisher or news agency has explicitly stated, 'Do not crawl me or crawl me for any reason other than indexing,' so that others people can find this content," he said. explain. "But this gray area. And I think it will pass the court. 

This is a little to say. While Suleiman’s comments certainly seem to anger content creators, he’s not entirely wrong. It’s unclear where the legal boundaries are regarding the training and output of AI models.

Most people who post content online as individuals are violating their rights in some way by agreeing to the terms of use offered by major social media platforms. Reddit's decision to license user posts to OpenAI wouldn't have happened if the social media giant believed its users had legal rights to their memes and manifestos.

The fact that OpenAI and other AI modeling companies have signed content deals with major publishers shows that big tech deals can be negotiated if you have a strong brand, deep pockets and legal team. In other words, anyone who creates content and posts it on the internet is creating free software unless they can hire or find lawyers willing to challenge Microsoft and its ilk.

In an article published by SSRN last month, Frank Pasquale, a law professor at Cornell Tech and Cornell Law School in the US, and Haochen Sun, an associate professor of law at the University of Hong Kong, explore the legal uncertainty surrounding the use of copyrighted data for AI training and whether courts would find such use fair. They conclude that artificial intelligence needs to be addressed at the policy level, as current law is inadequate to answer the questions that currently need to be addressed.

"Given significant uncertainties about the lawfulness of AI providers' use of copyrighted works, lawmakers will need to articulate a bold new vision to rebalance rights and responsibilities, as they did with the rise of the Internet (leading to the Digital Millennium Copyright Act of 1998)," they say. The authors suggest that the unpaid collection in creative works threatens not only writers, composers, journalists, actors and other creative teachers, but also the AI ​​general, who will ultimately be hungry for educational data. People will cease to do the work available on the Internet, they predict whether it gets used to the power of AI models, which reduces the marginal costs of content creation, zero and deprived of creators of the possibility of any reward. It is a future slimman. "The information economy changes fundamentally, because the cost of knowledge production can be reduced by zero costs at the maximum cost," he said.