Microsoft AI executive Mustafa Suleyman mistakenly believes that once content is published on the web, it becomes "freeware" available for anyone to copy and use without restrictions. When asked if AI companies were stealing intellectual property, he defended the practice, stating that content on the open web has long been considered fair use and freely available for replication. However, this assertion is incorrect and has led to Microsoft facing lawsuits accusing the company, along with OpenAI, of unlawfully using copyrighted online material to train their AI models. It is surprising to see a Microsoft executive publicly defending this position, as it is clearly incorrect. Although I am not a legal expert, it is important to note that copyright protection is automatically granted to creative works in the US as soon as they are created. This protection does not disappear when a work is published on the web; in fact, the difficulty in voluntarily waiving these rights has led to the development of specific web licenses by legal professionals. Furthermore, fair use is not based on a "social contract, " but rather a legal defense upheld by courts.

It allows certain uses of copyrighted material after considering factors such as the purpose and extent of the copying, and whether it harms the copyright owner. While some AI companies may argue that training on copyrighted content falls under fair use, few have been as audacious as Suleyman in publicly stating this viewpoint. In addition, Suleyman mentions the idea of using a robots. txt file to prevent scraping of websites. He acknowledges that explicitly stating not to scrape or crawl a website except for indexing purposes is a gray area that will likely be resolved through legal proceedings. However, it's worth noting that robots. txt is not a legally binding document and serves as a social understanding rather than a legal framework. Unfortunately, even this guideline is being ignored by some AI companies, including Microsoft's partner, OpenAI.