Microsoft has updated its terms of service to make clear that Copilot outputs are intended for entertainment purposes only and should not be relied upon for critical tasks.
The software giant’s legal documents now explicitly state that Copilot’s responses may contain errors, omissions or harmful content. "You must not use the output from the services or the services themselves for purposes that require factual outputs or outputs free of harmful content," the terms read.
This stance aligns with warnings from AI industry leaders. Earlier this year, OpenAI CEO Sam Altman acknowledged in a Senate hearing that AI models can produce inaccurate information. Microsoft’s clarification follows similar language in Google’s terms for its AI tools, which also caution users against trusting outputs for factual decisions.
The company added that users bear full responsibility for verifying Copilot’s outputs before acting on them. "Microsoft does not warrant that any output is accurate, complete or useful," the terms state. The policy update comes as Copilot is integrated into Microsoft 365 applications, including Word, Excel and Outlook.
Legal experts note that such disclaimers shift liability away from tech companies. "Terms like these are standard now," said Oslo-based tech law professor Lars Berg. "They protect providers when AI makes mistakes that cause real-world harm."
Source: techcrunch.com