In the current era, blooming of artificial intelligence (AI) is revolutionary. AI- driven technologies like financial forecasting, AI powered healthcare, innovations in biotechnology, nano technology, renewable energy technology, IoT driven sustainability and large language learning models (LLM) like GPT-4, Gemini to autonomous systems are reshaping how businesses operates and communication with people and technology. The rise of generative AI has ignited evolution in software development, content creation and even in creative arts which further makes AI more accessible to everyone and it has been integrated into mainstream applications like natural language processing (NLP), computer vision and Machine learning (ML).
The rapid growth in AI initiated the evolution of AI based startups which led to create intense competition in innovation and also created a havoc for more legal battles over the intellectual property rights. As technologies are getting more advanced everyday AI based companies begin to file patents aggressively in order to protect their innovations. This has led to legal war for patents. For example, the dispute between the Google’s waymo and Uber for innovation in self-driving technology, where Uber was required to pay 245 million dollars to settle a lawsuit concerning misappropriated trade secrets. Similarly in 2023, stability AI has faced legal challenges from another company Getty images for training its AI model with their copyrighted photographs without their permission. From these examples, it is clearly understood that there is growing importance to create awareness on Freedom to operate (FTO) analysis which facilitates the AI startups to explore the complex patent landscape and avoid costly legal disputes.
Why AI Startups must Prioritize Patent Clearance
Patent clearance becomes essential for AI based startups. Since the evolution of AI technology, unlike traditional software, innovations and inventions in AI often integrate with multiple technologies which makes them more prone to patent overlapping and risk for infringement. A single AI product might involve multiple patented components like machine learning models, data processing techniques and user interfaces which may be protected in the existing patents owned by the competitors. AI based startups risks unintentionally for infringing on third-party intellectual property rights which may lead to fines, expensive litigation and restrictions on the distribution of products. For example, in 2017, Deep mind faced scrutiny for its AI integrated health app for the usage of data rights which becomes an eye opener for many company to give importance for the thorough analysis of IP assessments. Clearing the patent directly impact the AI startups ability to attract investments from investors and venture capitalists which can lead to legal risks and uncertainty.
A well-documented freedom to operate (FTO) analysis provides an idea for the investors about the company that has taken proactive steps to minimize risks and ensure long term viability. Startups may struggle to secure more funds or high yielding partnerships without completing FTO analysis because stakeholders may fear for the risk of litigation. As regulations of AI is given importance globally, patent compliance is becoming even more critical. A strong FTO strategy ensures that the product is compliant with regional patent laws and licensing requirements particularly for the startups which aim to operate globally. AI startups can avoid legal battles, enhance credibility and establish a competitive advantage by integrating patent clearance for their invention and innovation. Some important patent clearance strategies for AI based startups are discussed below.
Patent landscape monitoring
AI based startups should continuously monitor the patent landscape from the idea stage, thereby it is not necessary to wait until the development of a product is complete. As AI is rapidly evolving field, it is crucial to stay updated on abandoned patents, new filings or licensing opportunities.
How to Implement?
- AI driven patent search tools plays a vital role to identify that the innovation has already been claimed. Some of the tools like Derwent innovation, Patent scout or the lens to track is very important to identify works in real time.
- Review patent classification codes relevant to AI like G06N in regular interval of time.
- Conducting the quarterly or biannual patent audits which is essential for the publication of new patent in the way that existing invention do not cause a great threat for FTO status.
Example:
An AI startup is developing an autonomous drone navigation software which notices that a competitor has already filed a similar patent covering path optimization method. By identifying this early, the startup can either modify the algorithm to design or explore the licensing options or update the product before the launch of the product.
Why it matters?
Monitoring the patents reduces the risk of unexpected legal disputes which also facilitates the startups to make informed decision about product development, strategic partnerships and potential acquisition based on the published patent.
Innovate smartly: Avoiding Patent Infringement
If an AI based startups identifies an existing patent document that could possibly create an infringement risk, it can create a new design around the existing patent by creating modifications in the technology. This can be done through re-engineering specific components in the existing technology or modifying or using the alternative approaches that do not overlap with patent claims.
How to Implement?
- Focusing on the patent claims is very important. First conduct the detailed analysis of patent claims and second focus specifically on the independent claims as it describes about the invention.
- Third, collaborate with subject matter experts and legal experts to explore the alternative methods.
- Finally, document all the modifications and technical differentiators for the future references to publish new patent.
Example
Assume, a startup is working on AI integrated predictive analytics for healthcare, identifies that a multinational company is holding patent on deep learning based anomaly detection method, instead of infringing, the company can switch to different methods like reinforcement learning or develop a new hybrid AI model which achieves the same outcome but with different approach.
Why it matters?
Designing around the patents facilitates AI startups to maintain FTO without spending for legal battles or expensive licensing fees, thereby promoting innovation without any single barrier.
Harness Open source and Patent Pooling Strategies for growth
AI based startups can strategically use the Open source based AI models to participate in patent pools in order to gain access for the existing patented technologies and reduce the risk of legal battles. Open source licensing should be used in an appropriate manner which provides legal framework for startups to develop AI applications without violating any aspect of existing patent document.
How to Implement?
- AI based frameworks must be utilized appropriately with permissive licenses. For example Tensor flow under Apache2.0 and Py torch under BSD.
- Check and analyze AI patent pools and cross licensing agreement to avoid the risk of litigation. For example LOT network.
- Strictly adhere to open source licensing terms in order to avoid unforeseen legal complications.
Example
Google’s Tensor flow is widely used for the development of AI model because of its open source licensing model Apache 2.0 which permits modification and distributes without claiming any royalty payments. An AI startup focusing on natural language processing (NLP) can be developed on tensor flow rather than developing a new proprietary model, thereby avoiding the potential infringement.
Why it Matters?
Leveraging open source models facilitated AI based startups to reduce developmental costs, minimize legal risk and accelerate innovation by developing over the existing technologies.
Secure Licensing Agreements and Early Patent Partnerships
If a startup identifies, that technology integrated in the existing patent by another inventor is the most important aspect for their product, then instead of claiming through legal battles, they can negotiate for early licensing agreement or collaborating with the patent holder for a strategic partnership to ensure smooth market entry. Instead of considering patent as a barrier for their innovation, AI based startup companies can consider it as an opportunity for making new collaborations.
How to Implement?
- Identify the important and specific patents which align with your AI product.
- Approach one who holds the patent to get license early or formal agreement to commit for joint ventures and research and development collaborations.
- Cross licensing agreement is helpful for both the parties as it is delivered through shared IP.
Example
AI based startup Wayve, focused on self-driving technology, thereby it further initiated the partnership with Microsoft to gain access for their Azure AI super computing capabilities. This strategic partnership helped wayve to leverage Microsoft’s IP when securing access to cutting edge computing infrastructure.
Why it Matters?
Ensure to receive licenses early, in order to prevent legal disputes, foster industry collaboration and allow startup companies to focus more on innovation rather than litigation.
Defensive publishing to Prevent future patent claims
Startup companies mainly focuses only on patenting their own innovations but defensive publishing is also an equally powerful strategy in which deliberately disclose the invention to the public claiming for the authority and also prevent the competitors from patenting the same invention. Defensive publishing safeguard the inventions from only restrictive patent claims which gives way to develop providing freedom to innovate.
How to Implement?
- Publishing technical papers, whitepapers, or open source documentation through which details of the AI methods, approach, algorithm or architecture can be revealed to the public.
- Submitting the innovations in prior art databases such as IP.com, prior art database or MIT’s technical disclosure commons.
- Ensure publications are incorporated with detailed technical descriptions, to qualify as prior art as well as to block future patent filings on similar technology.
Example
In 2014, Tesla open-sourced its electric vehicle patents not only to encourage more innovations but also to prevent competitors from claiming key technology.
Why it Matters?
Defensive publishing blocks patent trolls by securing exclusive rights over innovations in AI. This method enables startups to work freely without fear of infringement of claims on their disclosed technology.
From Patent landscape monitoring, Innovate smartly: Avoiding Patent Infringement, Harness Open source and Patent Pooling Strategies for growth, Secure Licensing Agreements and Early Patent Partnerships and Defensive publishing to prevent future patent claims, startups have multiple avenues to ensure freedom to operate (FTO) analysis while fostering innovation.
By adopting these five strategies AI based startups can protect intellectual property rights, mitigate legal risk and establish a strong foothold in their respective domains. As AI continues to bloom in the current era, one who approach patent clearance as a strategy will lead the future of AI innovation.
Author- Dr. S.Sarayu Priyadharshini, in case of any queries please contact/write back to us via email to chhavi@khuranaandkhurana.com or at Khurana & Khurana, Advocates and IP Attorney.
References
- (n.d.). Revolutionizing patent risk assessment: The value and practical implications of the Evalify’s infringement by usage. Evalify AI. Retrieved from https://www.evalify.ai/blog/revolutionizing-patent-risk-assessment-the-value-and-practical-implications-of-the-evalifys-infringement-by-usage
- Foley & Lardner LLP. (2022, October). 5 key IP considerations for AI startups. Foley & Lardner LLP. Retrieved from https://www.foley.com/insights/publications/2022/10/5-key-ip-considerations-ai-startups/
- The National Law Review. (n.d.). Best practices for developing winning IP strategies for AI companies. The National Law Review. Retrieved from https://natlawreview.com/article/best-practices-developing-winning-ip-strategies-ai-companies
- com. (n.d.). Defensive publishing and innovation protection strategies. IP.com. Retrieved from https://ip.com/innovation-power-suite/defensive-publishing/
- (n.d.). Patent strategy and administration: Defensive publication. Questel. Retrieved from https://www.questel.com/patent/patent-strategy-and-administration/defensive-publication/
- Hern, A. (2017, May 16). Google DeepMind 1.6m patient record deal was ‘inappropriate’. The Guardian. Retrieved from https://www.theguardian.com/technology/2017/may/16/google-deepmind-16m-patient-record-deal-inappropriate-data-guardian-royal-free