“Don’t be afraid, Microsoft will have you covered. » This is, in essence, the message sent by the company founded by Bill Gates to users of the artificial intelligence (AI) tools integrated into its office software (Office 365, including Word, Excel, Teams or PowerPoint) and help with writing computer code (Github Copilot). On Thursday, September 7, Microsoft agreed to pay all costs and damages to its customers in the event of a conviction for copyright infringement related to content generated by one of its AI software.
“We will take responsibility for potential legal risks,” promises legal director Hossein Nowbar in a blog post.
Microsoft has become a major player in AI, by investing in OpenAI, the creator of the chatbot ChatGPT, and by forming partnerships with the Californian start-up to use its numerous AI software capable of generating text and images.
Deploy AI in your own software
To be profitable, its strategy consists in particular of marketing these tools to client companies via its offer of online services in the cloud, but also by deploying OpenAI’s AI in its own software: integration into its famous Office 365 suite. , widely used by businesses, is currently in the testing phase. Its code writing istance tool is already very frequently used by computer scientists, for 19 dollars per month (17.74 euros).
But the adoption of these offers by large companies is hindered by the legal risk that weighs on artificial intelligence in the field of copyright. Many content creators and rights holders believe that large artificial intelligence models have been trained on vast quantities of images, books, articles and copyrighted works, without permission. Computer coders have filed a complaint against Github, the image bank Getty has attacked the AI manufacturer Stable Diffusion, media are planning to sue AI players, etc.
Uncertain and decisive standoff
Microsoft is well aware of this problem: “Some customers are concerned about the risk of copyright infringement claims if they use AI-generated content. This is understandable, given recent complaints from artists and authors about the use of their work in services using AI models,” recognizes the company.
You have 22.98% of this article left to read. The following is for subscribers only.