Microsoft has updated its Services Agreement to address content generated by AI. These changes come amidst an increase in AI-related language in platforms’ terms of service. Companies should be aware of the limitations placed on AI services, particularly in cases where, as with Microsoft, such limitations are not negotiated but rather exist in online terms.
The changes to the Services Agreement include:
- Definitions: The definition of content controlled by the Services Agreement has been expanded to include content generated by users via Microsoft’s AI services.
- Conduct Limitations: Users are prohibited from certain conduct associated with AI, including “jailbreaking” an AI system and generating intentionally deceptive content.
- Service-Specific Terms:The Services Agreement added a new “AI Services” section to address AI-specific obligations.
- Reverse Engineering: Users may not use AI Services to reverse engineer models, algorithms, and systems;
- Data Extraction: Unless specifically permitted, users may not extract data from AI Services, including through web scraping, web harvesting, or web data extraction methods;
- Data Limitation: Users may not use AI Services, or data from AI Services, to create, train, or improve any other AI service;
- Content Use: Microsoft is permitted to process and store inputs into, and outputs from, the AI Services for purposes of monitoring and preventing abuse or harm;
- Third Party Claims: Users are solely responsible for responding to any third-party claims regarding the use of AI Services, including copyright infringement claims.
The updates to the Services Agreement will be effective on September 30, 2023.