Copyright indemnification for AI users- Bane or a boon

Generative AI technology is rapidly expanding across various platforms and applications, but this growth has also led to a surge in copyright disputes for developers. Recently, post the launch of its own conversational chatbot, Bard, Google faced a class action lawsuit alleging web scraping, including copyrighted content, for training purposes. Similarly, ChatGPT, another popular AI chatbot, received criticism from media and authors, leading to multiple class action suits, notably one by the Authors Guild in the US. As a remedial step to these challenges, Generative AI companies are implementing Copyright Indemnification Policies to safeguard their users from potential copyright lawsuits stemming from the use of their Generative AI technology. These policies, also known as copyright infringement insurance, aim to protect individuals or entities from liabilities arising from claims of copyright infringement. This article shall delve into the real time implementation and extent of protection offered under copyright indemnity policies.

Copyright concerns with Generative AI

Generative AI raises copyright infringement concerns in several aspects. Primarily, there's the issue of AI-generated content or the AI itself possibly infringing on copyright licenses. The eligibility of AI-generated works for copyright protection sparks debate, as copyright laws traditionally protect only human-authored creations. Some argue that AI-generated content lacks human creativity and thus doesn't meet copyright criteria, while others suggest that the creators of AI models generating such content should be credited. Another concern arises from the use of copyrighted data to train AI models, also termed as training data. AI models utilize copyrighted media like text, images, and music for training purposes. In defense of such practice, it is contended that using copyrighted data for training AI models constitutes as fair use, while others argue it constitutes infringement. This issue plays a key role in the ongoing dispute between Getty images and Stability AI, wherein Getty Images has alleged that Stability AI unauthorizedly used 12 million images from Getty’s database to train its Stable Diffusion AI art generator.

The creation of unauthorized derivative works using Generative AI forms is also a cause for concern. Derivative works like book translations, art reproductions etc, are based on existing copyrighted works. Creation of such works through generative AI without the consent of the owner of the copyright would inevitably expose the user to copyright infringement claims.

With the view to tackle these pertinent issues, big tech generative AI providers are offering indemnification against potential copyright claims through indemnification policies. However, effectiveness of such policies in quelling infringement risks altogether, remains a subject of scrutiny.

Gen-AI providers offering indemnification- the methods, risks and reasons

With rising skepticism around use of AI, key players in the industry have introduced policies ensuring a copyright shield for its users. Microsoft has implemented ‘Copilot Copyright Commitment’, wherein it has pledged to assume responsibility for potential liabilities arising from use of Microsoft’s AI Copilot services and its outputs. Further, Microsoft shall defend its users from any third party claims against such use. Getty Images offers complete legal and financial indemnity to the businesses using its image generating AI tool against copyright claims. Adobe has begun offering IP indemnity for commercial users of Firefly, its generative AI-powered image creator application. Shutterstock has also introduced an indemnity policy for its enterprise-level customers, covering the use of AI-generated images created on its platform. Open AI, also offers copyright indemnity to enterprise-level ChatGPT customers and users of OpenAI’s developer platform. Open AI has also ventured into introducing built-in copyright considerations on the ChatGPT platform. The key motivator for providers to offer indemnities to users is to provide assurance of risk-free exploitation of AI. With rising skepticism around use of AI, providers aim to instill greater confidence in users to utilize their generative AI tools. The increase in infringement claims against AI users has deterred the potential users. Thus, by providing appropriate indemnities, providers seek to reassure users and encourage secure use of AI tools. Additionally, with leading AI companies backing its users, it shall dehort claimants from initiating infringement claims.

In principle, the indemnities offered by companies is a positive step for users, they should not be regarded as a cure-all solution to all potential IP infringement claims. The indemnities are usually accompanied by limitations and exclusions. For instance, the Microsoft’s customer Copyright Commitment is formulated to defend the customer against third party IP claims wever, this indemnity is subject to various conditions, including:

Microsoft is given prompt notice of the claim and controls its defense;
The product must be paid for;
It must be used within license scope, unmodified and not combined with anything else;
Microsoft can terminate the license and refund fees paid if it can’t settle the claim.

Thus, indemnities may usually be capped with limitations and may not be available to anyone and everyone for all situations. Hence, an oblivious reliance on these policies may not bode well for the user. Regardless, the ability to generate high-quality content without the threat of lawsuits can be a game-changer for users, especially startups and smaller enterprises in general that don't have the legal muscle to fight lengthy copyright cases.

Conclusion

In the past, the AI providers have been known to disclaim responsibility for any infringement arising from the output of their AI platforms. Nonetheless, the efforts of the generative AI providers to assure risk free use of AI is surely a welcoming move and has brought a significant shift in the ecosystem. However, the offering of indemnities may also add fuel to the already ongoing dispute between the AI providers and copyright industries over use of copyrighted material.

What remains abundantly clear is that advancements in this space continue to accelerate and offering copyright indemnification to users may not be completely unconditional, but shall be a critical consideration for AI users seeking to leverage AI technologies while managing potential legal risks.

References: