Please do not use our content.However, your content will use

• Microsoft-backed OpenAI, Google, and Anthropic have banned their content from being used to train other AI models.
• However, these companies use other online content for their own model training.
• Can Big Tech do both? Reddit and others are trying to stop this.

In the new era of generative AI, tech giants are following a “do what I say, not what I do” strategy when it comes to consuming online content.

Microsoft-backed OpenAI collaborates with Google, and Google-backed Anthropic has been training generative AI models using online content created by companies for years. This was done without asking for special permission, and is part of a legal battle to determine the future of the web and how copyright law will apply in this new world.

The tech industry will probably argue that their approach is fair use. That has not yet been decided. However, these big tech companies do not allow their content to be used to train other AI models. So why are they allowed to do this to others?

Let’s take a look at the terms of service for Claude, Anthropic’s AI assistant.

“You may not access or use the Service in any of the following ways, and if any of these restrictions are inconsistent with, or ambiguous in connection with, the Terms of Use, Developing products or services that compete with our services, including developing or training artificial intelligence or machine learning algorithms or models.”

Below is an excerpt from the beginning of Google’s Generated AI Terms of Service.

“You may not use the Services to develop machine learning models or related technologies.”

Here are the relevant sections of the OpenAI Terms of Service: This is the company behind ChatGPT.

“The output from the service cannot be used to develop models that compete with OpenAI.”

These companies are not stupid, but they are hypocritical

These companies are not stupid. They know that high-quality content is essential for training new AI models. So it stands to reason they don’t allow the output to be used in this way.

But why would other websites and companies make their content freely available for use by these tech giants to train their models?

Insiders reached out to OpenAI, Google and Anthropic on Friday for comment. At the time of publication, they had not responded.

Reddit and Others Claim Enough is Enough

Other companies are just starting to realize what’s going on, but aren’t satisfied. Long used to train AI models, Reddit will start charging for access to its data.

Reddit CEO Steve Huffman said, “Reddit’s data corpus is really valuable, but you don’t have to give all that value away for free to the world’s biggest companies.”

In April, Elon Musk accused OpenAI’s main backer, Microsoft, of illegally using Twitter data to train an AI model. “It’s lawsuit time,” he tweeted.

“There are so many flaws in this premise that we don’t even know where to start,” a Microsoft spokesperson wrote in an email to Insider when asked for comment.

OpenAI CEO Sam Altman is trying to be more thoughtful about the issue by working to develop new AI models that respect copyright. According to Axios, he recently said that he is “working on a new model where AI systems are rewarded for using your content or using your style.”

The publisher, including the insider who created this article, has a vested interest here. Some publishers, including News Corp., are already asking tech companies to pay to use their content to train AI models.

Current AI model training methods are ‘breaking’ the web

One former Microsoft executive thinks there’s something wrong here. Steven Sinofsky recently said that current methods of training AI models are “breaking” his web.

“In the past, crawling was allowed in exchange for clicks, but now crawling simply trains models and provides no value to authors and copyright holders.” he says. tweeted. An insider reached out to him for comment, but he could not respond because he was traveling on Friday.