Anthropic’s Updated Terms of Service: Enhanced Copyright Protection for AI Outputs

Anthropic, a leading generative artificial intelligence company, recently updated its commercial terms of service, effective January 1, 2024, to address important concerns regarding intellectual property and data usage. This update is especially important for users of Anthropic’s Claude API, which is also available through Amazon’s generative AI development suite, Bedrock.

Under the new terms, Anthropic has taken a strong position to protect its customers from copyright infringement claims related to the permitted use of their services. The company is committed to indemnifying its Claude API enterprise customers against such claims, promising to defend itself against claims that the customer’s paid use of Anthropic’s services, including data used to train their models, infringes third-party intellectual property rights parties, such as patents, trade secrets, trademarks or copyrights.

The move puts Anthropic alongside other major generative AI vendors such as Microsoft, Adobe, Shutterstock, OpenAI, IBM and Google, which have also implemented similar intellectual property protection measures for their generative AI outputs. However, these protections come with certain limitations. For example, the indemnity does not cover claims related to customer prompting or use of the service that violates the terms of use, arise from willful misconduct or violations of law, modifications made by the customer to the services or results, or the combination of services or outputs with technology not provided by Anthropic. In addition, certain patent or trademark infringements are also excluded from this protection.

In addition, it is important for businesses considering the use of generative AI tools to thoroughly analyze the terms of the vendor, taking into account not only legal factors, but also non-legal considerations such as pricing and technical capabilities. This comprehensive review should include a review of intellectual property protection provisions and any relevant exemptions offered by the provider, particularly for paid or enterprise customers. Such diligence is critical in the current environment, where copyright and privacy lawsuits are still pending against generative AI vendors related to the deletion of copyrighted works for training AI models. These legal battles present unresolved intellectual property issues, making the protections offered by companies like Anthropic a welcome addition for their clients. However, it is vital for customers to fully understand and consider these terms and indemnification provisions in the context of their specific or potential uses​​.

Legal actions such as those brought by Universal Music Group against Anthropic in October 2023 and the lawsuit against OpenAI and Microsoft by author Julian Sancton highlight the complexity and evolving nature of copyright law in the age of artificial intelligence. These cases highlight the importance of AI companies proactively addressing copyright issues and ensuring they have robust protections in place for their clients. The recent update to Anthropic’s terms of service is a step toward greater clarity and certainty in this rapidly advancing field.

Image source: Shutterstock

Leave a Comment