Microsoft says Copilot isn't just 'for entertainment purposes' after its terms of service language goes viral

• Microsoft said it will update Copilot's terms of use after they went viral.
• Users highlighted terms that said Copilot is "for entertainment purposes only."
• Microsoft said the phrasing "is legacy language" that dates back to when Copilot was part of Bing.

Microsoft is making sure its Copilot Terms of Use no longer steal the show.

The tech giant said it will be updating its user agreement after viral posts pointed out that Copilot said it is "for entertainment purposes only," a far cry from how Microsoft has sold its AI tool.

"The 'entertainment purposes' phrasing is legacy language from when Copilot originally launched as a search companion service in Bing," a Microsoft spokesperson said in a statement, first published by PCMag. "As the product has evolved, that language is no longer reflective of how Copilot is used today and will be altered with our next update."

In recent days, users on X have highlighted the terms, suggesting that Microsoft isn't confident in its flagship AI tool.

"Copilot is for entertainment purposes only," the terms read. "It can make mistakes, and it may not work as intended. Don't rely on Copilot for important advice. Use Copilot at your own risk."

In contrast, during the company's most recent earnings call in January, CEO Satya Nadella praised "Microsoft 365 Copilot's accuracy and latency powered by Work IQ," an intelligence tool within the AI agent.

It is worth noting that the Copilot agreement is not the only term of use that users must agree to. Users also agree to Microsoft's Services Agreement, which does not mention "entertainment purposes" in its AI subsections.

In contrast, OpenAI, Anthropic, xAI, and Meta do not use the term "entertainment purposes" in their user terms, though the overall scope of their agreements use similar language to avoid legal liability.

Previous Copilot Terms of Use contain references to "entertainment purposes" dating back to February 2023. In November 2023, Microsoft announced that to "make Copilot more accessible to everyone," Bing Chat and Bing Chat Enterprise were now Microsoft Copilot.

Shriram Krishnamurthi, a professor of computer science at Brown, wrote that odd terms of service are a feature, not a bug, for Microsoft.

Microsoft's wording stands out from competitors

OpenAI, Meta, Anthropic, and xAI take a very different approach from Microsoft, though all are explicit about the limits of their AI models.

For example, OpenAI's terms state, "You accept and agree that any use of outputs from our service is at your sole risk, and you will not rely on output as a sole source of truth or factual information, or as a substitute for professional advice."

Elon Musk's xAI, which was acquired by SpaceX in February, explicitly requires consumer users to indemnify the company, meaning they assume any legal liability for essentially all their use of the platform.

"To the fullest extent permitted by law, you will defend, indemnify, and hold xAI and our parents, subsidiaries and affiliates, and our and their respective agents, suppliers, licensors, employees, contractors, officers, and directors (collectively the "xAI Indemnitees") harmless from and against any and all claims, damages," xAI states in its terms.

Meta's AI terms for most users state that not only should users "Not rely upon outputs for any purpose or use outputs to inform professional advice or decisions related to medicine, finance, law, or pharmaceuticals," but that some of those exact scenarios would not be considered acceptable uses.

"Solicit professional advice (including but not limited to medical, psychological, financial, or legal advice) or content to be used for the purpose of engaging in other regulated activities (including but not limited to political campaigning or lobbying)," Meta states as examples of unacceptable uses in its terms.

Generative AI remains a relatively new technology, but lawsuits are already swirling over the extent to which AI companies can be held liable for the outputs of their chatbots.

OpenAI is facing a dozen lawsuits in California state court that allege GPT-4o, a now-retired AI model known for its sycophantic responses, is responsible for harm to users, including some like Matthew Raine who died by suicide.

Last month, Nippon Insurance Company sued OpenAI in federal court in Illinois, alleging that the AI startup was responsible for an insurance customer repeatedly causing legal headaches for the company after ChatGPT told the customer that her lawyer was gaslighting her about a settlement she had reached with Nippon.

OpenAI has said it wants the public to wait until all the facts in mental-health-related cases come to light and that it will continue to improve its models regardless. It has not commented on the Nippon case.