Threats posed by AI to Trade Secrets

Abstract

The concept of Artificial Intelligence is not a recent one it has been with us since the 1900s but it gained massive attention in the year of 1955 after John Mcarthy first used the term "Artificial Intelligence" at the Dartmouth Conference, and since then Artificial Intelligence has managed to take the world by storm and has transformed various industries. Reminiscing the words of Sir late Martin Luther King, “when god built a church the devil built a chapel”, the benefits of Artificial Intelligence is as much as the harm that it might cause. After being in the headlines for infringing the IP rights of creators and inventors, AI is now in the limelight for posing threats to Trade Secrets. Trade secrets are protected by confidentiality, i.e., the person who created the trade secret was the only person who knew about and they could use this knowledge to make money. However, with the advent of AI, it is becoming extremely difficult to keep trade secrets confidential.

Introduction

Since the Dartmouth Conference of 1955, Artificial Intelligence has permeated into every aspect of our lives, be it a student or a business leader Artificial Intelligence has its influence on each one of us. Over the years Artificial Intelligence has become so sophisticated that people are now referring to it as "Super Intelligent AI” and has become the new eye-catching tool in the tech world that everyone wants to have their hands on. Since the evolution of sophisticated Chatbots like ChatGPT, Google Bard, Co-Pilot, etc. the world has been fascinated with the technology, and its ability to produce results however it cannot be denied that this disruptive technology comes with its own sets of risks. In the last two years, people have formed divided opinions about the technology while some advocate for granting AI the status of authorship and inventorship, the rest are trying to make AI accountable for infringement of IP rights. While the courts are busy striking a balance between technology and law, the focus has now shifted from the topic of accountability, authorship, and inventorship to the threats posed by Artificial Intelligence to Trade Secrets. These tools pose a potential threat to a company's trade secrets, as an employee may accidentally disclose sensitive information by using generative AI applications. This has led companies like Samsung to ban its employees from using Artificial Intelligence tools for work after losing its trade secrets.

The history of trade secrets can be traced back to the Romans, where Romans used slavery to control the line of possessors who were aware of trade secrets¹. However, in modern times business Trade Secrets are protected by confidentiality, Non-disclosure Agreements, etc. However, with the advent of Artificial Intelligence, the rules have changed. While businesses and companies are trying to be open and are encouraging employees to use Artificial Intelligence to complete a task, ²they are also making their confidential information vulnerable leading to potential data leaks. The inputs provided to the AI tool processed for generating predictions and outcomes are retained by the system only to be reused and build up internal capabilities³. The AI tool is exposed to continuous learning, assimilating, and creating newer versions without recourse to honoring IP rights. In the process, the confidentiality aspect of the trade secret stands compromised as commodification of unique designs, innovation, and products almost become a norm de guerre leading to the destruction of intellectual value and rights of people and businesses. This uncalled-for erosion of worth will stifle enterprise and have cataclysmic consequences for the economy.

Intersection of Trade Secrets and AI: Potential Data Leaks

The relationship between Trade Secret and AI is a tricky one. According to WIPO trade secrets are “Intellectual Property rights in confidential information which may be sold or licensed”⁴ To safeguard trade secrets, the party getting the information is required to make an express promise⁵ either by signing an NDA, contract, etc., to keep the information confidential before learning the trade secret. However, this approach cannot happen when using a generative AI system as the party at the receiving end is not human. Artificial Intelligence is a Large Language Model (LLM) that is fed with huge chunks of information to generate results, sometimes this information provided to the system might contain sensitive information such as Trade secrets, crucial business confidential information, financial records, sensitive personal information, etc., ⁶Moreover, Artificial Intelligence systems like ChatGPT store, collect, and process large amounts of information including user prompts, therefore if not contained and filtered out properly there is a high chance of this information being leaked⁷. AI tools are capable of generating new content by ingesting lots of data, both from the public domain as well as information provided to it. This data is usually stored on the company servers that own the AI tool. However, this process has raised a wave of concerns and worries regarding confidential information being used by AI. After incidents of sensitive info getting out to others after using generative AI have been reported many businesses have either put a total ban or have limited the use of AI at work.⁸

Instances of AI data Leaks

Trade secrets are fragile assets, once lost they cannot be recovered. Integrating Trade Secrets as prompts in AI tools compromises confidentiality as the machine is incapable of filtering out what information is confidential and what is not.

• Samsung Trade Secret Leak:
  Last year tech giant Samsung's trade secret was leaked by its employees while using an Artificial Intelligence tool.⁹ It was reported that the company’s engineers while trying to fix the source code using the infamous AI tool ChatGPT gave out confidential information in the form of prompts to the AI tool leading to a data leak. 
• Sundstrom Case:
  In the very recent case of West Technology Group LLC et al v. Sundstrom¹⁰ , the Plaintiff sued its ex-employee in the U.S District Court for the District of Connecticut for leaking company trade secrets, the Plaintiff claimed that the Defendant has used the AI tool “Otter” to “record and transcribe confidential meetings, analyze sensitive data which included consumer data, pricing information, and proprietary manufacturing process” before exiting the Plaintiff’s company.

These incidents emphasize how essential it is to safeguard valuable intellectual property from being stolen in the face of cutting-edge innovations like artificial intelligence.

How can companies protect their Trade secrets from getting leaked?

While the courts are trying to deal with the menace caused by AI tools, companies can try to safeguard their trade secrets from being leaked by strategizing some mitigation steps on a grass-root level. A few steps that companies can implement are¹¹

• Blanket Ban
  One way in which companies can stop trade secrets from being leaked is by putting a complete ban on the use of AI tools for work-related projects.
• Access Control
  Companies can create policies to limit access to private data. They can also try to limit the number of employees who can have access to and interact with AI tools and examine the kinds of inputs provided by the employees when using AI tools.
• Creating Awareness and Employee Training
  Companies from time to time should update their employee handbooks, agreements, and procedures to address the usage of AI tools at work and also provide training sessions to educate their employees on how to manage confidential data and highlight the legal consequences that they can face for misappropriation of trade secrets.

Position in India

India does not yet have a codified law that specifically protects trade secrets, however, the Indian courts from time to time have discussed and upheld the importance of trade secrets in catena of cases.

In the case of Burlington Home Shopping Pvt. Ltd v. Rajnish Chibber¹² , the Plaintiff was a mail order service company the Plaintiff ran its business by publishing mail order catalogues dealing with several consumer items to a select list of clients, the Plaintiff had made a major investment in the compilation of a list of clientele/customers database. It was found by Plaintiff that a former employee of (Defendant) the company has posed himself as a competitor of the Plaintiff and has managed to get a copy of the secret database containing client information and is using it to establish a connection with the Plaintiff’s customer. The Hon’ble Delhi High Court in this case held that the “defendant have indulged into slavish imitation of plaintiff's compilation making out a clear case of infringement”

In another case of Markets and Markets Research Pvt Ltd v. Meticulous Market Research Pvt Ltd and Ors ¹³, the Plaintiff was engaged in the business of providing specialized market research and intelligence to businesses and clients. It was contended by the Plaintiff that they use confidential methods to create reports that cater to specific markets, which were bought by high-profile clients. In this case, the Defendant (2 to 7) were former employees of the Plaintiff who have joined Defendant (no 1) The Plaintiff alleged that the Defendants were copying the format and content of the Plaintiff's market reports to make their own. Suspecting theft of confidential data, the Plaintiff had also filed a criminal complaint against the Defendant. Taking swift action the Hon’ble Delhi High Court held that the “Defendants action infringed the rights of the Plaintiff and passed an injunction order in favour of the Plaintiff”

This ruling was also maintained in the case of Indian Explosives Pvt Ltd v. Ideal Detonators Pvt Ltd and Ors ¹⁴, the Plaintiff was engaged in the business of manufacturing industrial explosives and shock tube explosives. It was claimed by the Plaintiff that the shock tubes cannot be copied without replicating or copying the drawings in two dimensions first. It was further claimed that former employees of the Plaintiff had stolen confidential information, trade secrets, and drawings related to shock tube detonators during the course of their employment and have sold them to Ideal Detonators (Respondent) for profit. The Hon’ble Calcutta High Court was of the opinion that “there is an involvement of the Respondent in setting up of a rival plant and the unauthorized use of drawings and other documents of the plaintiff and the confidential information of the plaintiff particulars” and passed an order in favour of Plaintiff restraining the Respondents from transferring, disclosing the information further.

Similarly in the case of Birla Corporation Ltd. v. Adventz Investment & Holdings Ltd ¹⁵, the Hon’ble Supreme Court held that “Information contained in a document if replicated, can be the subject of theft and can result in wrongful loss, even though the original document was only temporarily removed from its lawful custody for the purpose of extracting the information contained therein.”¹⁶

Even though the courts from time to time by their judge-made law have protected trade secrets and confidential information, there is still a requirement of a codified law. In this line, the Law Commission of India in its 289th Report has proposed "The Protection of Trade Secret Bill 2024”. Hon’ble Justice Ms. Prathiba M Singh of the Delhi High Court has also highlighted the need for a codified law protecting trade secrets. In her comment Hon’ble Justice commented "that in the current economic scenario, we have big innovations happening, especially within the Artificial Intelligence (AI), technology and start-up ecosystem she further stated that data is very vital to all these industries and one leak of any sensitive information can have a crippling impact on the progress of an entity, Even with big companies, a substantial capital is invested in research and most of the data that they seek to protect is at the research stage thus incapable of being protected under the system of patents and copyright, which may not afford adequate protection against the misappropriation of such data. Loss of such data has serious economic implications for such companies” ¹⁷.

Section 2(f) of the proposed bill in hand defines Trade Secret as “any information that is secret in the sense that it is not generally known among or readily accessible to persons, derives commercial value on account of being secret, has been subject to reasonable steps under the circumstances by the holder of such information to keep it secret and disclosure of such information is likely to cause damage to the holder of such information”. Section 7 of the proposed bill deals with relief that a court may grant to the Plaintiff in any suit for misappropriation of Trade Secrets, the reliefs include injunction, damages/accounts of profit, destruction of documents, objects, materials, etc., and pecuniary remedies.¹⁸

Before the proposed bill the Indian Courts in plethora of cases have taken the similar approach when granting reliefs to aggrieved parties who have suffered loss due to loss of trade secrets and confidentiality breach, however the current bill provides a more structured and comprehensive framework for protecting trade secrets in the face of emerging technologies like Artificial Intelligence.

Conclusion

While the world continues to grapple with the need to retain the sanctity of trade secrets in the context of AI applications, a combination of legal and regulatory initiatives can arrest the steady decline. Cautioning employees and advising them of discrete use to prevent inadvertent disclosures can be the first step. Another option could be to formulate codified laws and regulations that will act as a deterrent to mischief makers. A collective and conscientious approach in crafting a way forward is needed.

Authors: Vikrant Rana and Shilpi Saurav Sharan

Swayamsiddha Das, Associate at S.S. Rana & Co. has assisted in the research of this Article.

¹ Watson, A. (1996), Trade Secrets And Roman Law: The Myth Exploded, Tulane European& Civil Law Forum, 11(1996), pp 26
² Divyanshi Sharma, Software company employees use ChatGPT to write code, CEO says months of work can now be done in days, India Today,(Apr. 26 2024, 10:30 am) https://www.indiatoday.in/technology/news/story/software-company-employees-use-chatgpt-to-write-code-ceo-says-months-of-work-can-now-be-done-in-days-2383492-2023-05-24
³ Trade Secrets and Generative AI: Protective Measures In an Evolving Technological Landscape, Jonesday, (Apr.30, 2024, 11:00 am), https://www.jonesday.com/en/insights/2023/06/trade-secrets-and-generative-ai
⁴ Trade Secrets, WIPO, (Apr.26, 2024, 11:00 am), https://www.wipo.int/tradesecrets/en/
⁵ R. Mark Halligan, Artificial intelligence and trade secrets, (Apr. 26, 2024, 11:10 am), https://www.reuters.com/legal/legalindustry/artificial-intelligence-trade-secrets-2023-12-11/
⁶ David Stuart, Safeguarding Data Integrity and Privacy in the Age of AI-Powered Large Language Models (LLMs), Sentra, (Apr. 29, 2024, 10:00 am), https://www.sentra.io/blog/safeguarding-data-integrity-and-privacy-in-the-age-of-ai-powered-large-language-models-llms#:~:text=LLMs%20can%20inadvertently%20disclose%20sensitive,improper%20handling%20of%20user%20inputs
⁷ Data in Hot Water: The Cybersecurity Risks of Generative AI, Forcepoint, (Apr. 30, 2024, 2:00 pm), https://www.forcepoint.com/blog/insights/data-cybersecurity-risks-generative-ai
⁸ Id at 3
⁹ Priya Pathak, Samsung’s trade secrets have reportedly leaked because employees shared a bit too much info with ChatGPT, Financial Express, (Apr.26, 2024, 11:00 am), https://www.financialexpress.com/life/technology-samsungs-trade-secrets-have-reportedly-leaked-because-employees-shared-a-bit-too-much-info-with-chatgpt-3038108/
¹⁰ West Technology Group LLC et al v. Sundstrom, 3:2024cv00178
¹¹ Trade Secrets and Generative AI: Protective Measures In an Evolving Technological Landscape, Jonesday, (Apr.30, 2024, 11:00 am), https://www.jonesday.com/en/insights/2023/06/trade-secrets-and-generative-ai
¹² Burlington Home Shopping Pvt. Ltd v. Rajnish Chibber, 995(15)PTC 278(Del)
¹³ Dr. Sudipta Banerjee v. L.S. Davar & Company & Ors
¹⁴ Indian Explosives Pvt Ltd v. Ideal Detonators Pvt Ltd and Ors, IA NO. GA/1/2023
¹⁵ Birla Corporation Ltd. v. Adventz Investment & Holdings Ltd, AIR 2019 SUPREME COURT 2390
¹⁶ 289th Law Commission Report on Trade Secrets and Economic Espionage, March 2024

¹⁷ 289th Law Commission Report on Trade Secrets and Economic Espionage, March 2024
¹⁸ The Protection of Trade Secret Bill 2024