Microsoft Publishes First Responsible AI Transparency Report
Touted as the "first of its kind," Microsoft's Responsible AI Transparency Report was published on May 1, 2024.[1] In this report, Microsoft details its top concerns and risks posed by generative AI and the tools it employs to mitigate those risks.[2] This report follows a July 2023 meeting between President Biden and seven leading AI companies, where the White House secured voluntary commitments from these companies to promote "safe, secure, and transparent development of AI technology."[3]
Generative AI and Copilot
Generative AI refers to artificial intelligence models and applications capable of creating original content, including text, image, and audio. One of Microsoft's generative AI tools for consumers is Copilot.[4] Copilot is Microsoft's "everyday AI companion" that uses a conversational chat interface for text-based interaction.[5] You can use it to search for specific information, generate text formats (like emails or summaries), and even create images based on your text prompts.[6] While undeniably powerful, these tools, including Microsoft's Copilot, pose unique legal risks, such as potential exposure to intellectual property infringement for users.[7] Following the White House's initiative, Microsoft has taken steps to absorb some liability for potential infringement arising from using its tools.[8]
Microsoft's Customer Copyright Commitment
Through its Customer Copyright Commitment, Microsoft promises to defend and indemnify its consumers against third-party claims when using Microsoft's generative AI tools.[9] More specifically, it promises to "defend commercial customers sued by a third party for copyright infringement for using Azure OpenAI Service, [their] Copilots, or the outputs they generate, and pay any resulting adverse judgments or settlements, as long as the customer meets basic conditions such as not deliberately trying to generate infringing content and using [their] required guardrails and content filters" (emphasis added).[10] Microsoft's Chief Legal Officer, Brad Smith, provided three reasons for implementing this program: (1) Microsoft supports its customers, (2) it understands authors' copyright concerns, and (3) it has built guardrails into its Copilot programs to help respect authors' copyrights.[11] As Microsoft's Commitment was only expanded upon in late 2023, it has had little time to be tested in practice.
Uncertainties and Potential Applications
It remains to be seen how Microsoft will fulfill its Commitment, particularly regarding how it analyzes whether customers meet the "basic conditions." Additionally, Microsoft's use of the phrase "such as not attempting to generate infringing content" could be used later to deny benefits to customers. In other words, a user's intent in using generative AI may not always be clear.
The Commitment is silent on the criteria Microsoft would apply when deciding whether to defend and indemnify a customer suspected of intentionally creating infringing content. Despite these uncertainties, Microsoft's efforts to stand behind its products and consumers are noteworthy.
Looking Ahead
This report, especially the Customer Copyright Commitment, provides valuable insight into the evolving generative AI industry and its liability issues. Even without this commitment, companies like Microsoft might be held liable when their tools are used to create copyright-infringing materials, especially in cases of unintentional infringement. One might expect other leading AI companies to follow Microsoft's lead and develop similar commitments to demonstrate confidence in their product safeguards.
[1] Microsoft Corp., Responsible AI Transparency Report: How We Build, Support Our Customers, and Grow (2024).
[2] See id.
[3] Fact Sheet: Biden-Harris Administration Secures Voluntary Commitments from Leading Artificial Intelligence Companies to manage the Risks Posed by AI, WH.Gov (July 21, 2023), https://www.whitehouse.gov/briefing-room/statements-releases/2023/07/21/fact-sheet-biden-harris-administration-secures-voluntary-commitments-from-leading-artificial-intelligence-companies-to-manage-the-risks-posed-by-ai/.
[4] Yusuf Mehdi, Announcing Microsoft Copilot, Your Everyday AI Companian, Microsoft (Sept. 21, 2023), https://blogs.microsoft.com/blog/2023/09/21/announcing-microsoft-copilot-your-everyday-ai-companion/; see also Dan Wahlin, et al., Adopt, Extend, and Build Copilot Experiences Across the Microsoft Cloud, Microsoft (Dec. 20, 2023), https://learn.microsoft.com/en-us/microsoft-cloud/dev/copilot/overview.
[5] Id.
[6] See id.
[7] See Brad Smith, Microsoft Announces New Copilot Copyright Commitment for Customers, Microsoft (Sept. 7, 2023), https://blogs.microsoft.com/on-the-issues/2023/09/07/copilot-copyright-commitment-ai-legal-concerns/.
[8] Id.
[9] Supra at note 1.
[10] Id.
[11] Supra at note 7.
