When you are a start-up company, your IP (Intellectual property) is the most important (and probably only) asset. New generative-AI tools are becoming more advanced and efficient and are changing your R&D team workflow, allowing them to do more (coding) with less (time and people).
However, using generative-AI tools in the development of your IP is also risky. There are uncertainties relating to ownership and protection of the generated content, as well as risks relating to data leakage and data protection, that you should be aware of. Some of the key risks are:
• Ownership challenges. It is not yet clear whether software entirely created with Generative-AI tools is protected under copyright, mainly since it is not created by a natural person. This means that you probably would not want your most valuable code to be developed using generative-AI tools. For your clients and investors it is fundamental that you will be able to substantiate explicit ownership or license of the IP that is used in your product.
• Patent protection. Patents created by a generative-AI tool will likely not be registrable in many jurisdictions under current legislation. In fact, the Israeli Registrar of Patents recently rejected an application which specifically indicated that the inventor was not a natural person but rather a machine. Clearly, the legislation will need to be updated to reflect these technological advancements, but this could take several years.
• Excessive rights to re-use your data. Keep in mind that your use of generative-AI platforms is always subject to the terms of service of the platform you are using, which usually reserves the right to use the content you upload to the platform to train their models and tools and improve their service. This means that the generative-AI platforms have access to your data, a license to use it and publicly display it.
• Confidentiality breach. Other users of the generative-AI tools may receive similar output, and even output that contains your data (and in such case, they may claim ownership to such content). Your employees may provide confidential information of the organization which may be re-used and processed through the generative-AI tools. You do not want your confidential information or other sensitive data to be uploaded into a generative-AI platform by any of your employees. For these reasons, some of the biggest technology corporations have already adopted policies that restrict the use of generative-AI tools in the development of their products or the training of their AI models.
• Exposure to copyright infringement claims. Even where the terms of service of the generative-AI platform provide that it will not claim ownership in the output generated by the tools, this does not grant you full protection from third party claims relating to copyright infringement.
• Better have a corporate license. Make sure to obtain a corporate license to use the generative-AI tools and do not rely on personal or freemium licenses of your employees or vendors (even when they have signed an IP assignment agreement to your benefit), as in some generative-AI platforms personal users may not use the generated content for commercial use.
• Obtain a license to use the dataset. If you use an AI database to train your own AI tools, you should make sure you are allowed to do that, and whether the data in the platform you are using is licensed in itself. It is worth mentioning that the Israeli Ministry of Justice published in December 2022 an opinion which concluded that generally, the use of copyrighted materials in a ML (machine learning) context is permitted under existing copyright doctrines such as the "fair use" doctrine, the "incidental use" doctrine and even the "transient use" doctrine. Note however, that such protections may not always be available or applicable, specifically where commercial use is involved.
• Data Privacy Concerns. Generative-AI platforms may process and store user and personal data. It is essential to assess the service provider's data handling practices and ensure compliance with privacy regulations. There is a risk of inadvertent exposure or mishandling of PII if not properly controlled. Therefore, prior to engaging with an external generative-AI platform make a privacy assessment with respect to applicable data protection laws.
• No warranties to the generated content. Generative-AI platforms do not guarantee the accuracy, integrity or quality of the generated content and generally do not provide any warranty. Further, AI services may generate responses based on pre-existing data, potentially leading to biased, discriminative, or inaccurate information. Therefore you must verify that the generated content complies with the standard warranty you have committed to provide your customers with, and monitor the content.
• Due diligence aspects. The use of generative-AI tools is becoming an important item in due diligences exercise of transactional lawyers as part of M&A and investment transactions, so aside from being aware of the associated risks, the best thing you can probably do in order to be ready for a due diligence is to monitor the use of generative-AI in your organization and record all activities in this respect, in many ways similar to how you (should) monitor the use of OSS (Open Source Software).
Clearly this is an evolving field, and considering the above risks and the complexity of the technologies involved, you must consider adopting a responsible AI policy in your organization to minimize the risks and uncertainties of using generative-AI tools in your R&D activities.
(and no, we did not use generative-AI to write this article J)
This update is intended to serve as a general overview and does not constitute a legal advice or replacement for legal advise.
If you have any additional question, please contact us.
.png?cache=1686054550390)
