Legal Pitfalls of Using Generative AI to Draft Business Documents

“In the rapidly evolving landscape of technology and business, the utilization of generative artificial intelligence (AI) systems has surged, offering remarkable advancements and efficiencies across various sectors. One area where generative AI has gained significant attention is in the realm of drafting business documents. These sophisticated algorithms can autonomously generate contracts, agreements, and legal documents, promising time-saving convenience and accuracy. However, as the boundaries of AI implementation continue to expand, so do the legal complexities and potential pitfalls associated with its use. In this article, we delve into the legal challenges and risks that arise when employing generative AI systems to draft business documents, exploring the need for cautious navigation in this transformative era of automation and the law.”

Fittingly, the introductory paragraph above was generated by ChatGPT, an AI chatbot developed by OpenAI, in response to the prompt “Write me an introductory paragraph for an article titled “Legal Pitfalls of Using Generative AI to Draft Business Documents”. It demonstrates the power and potential that generative AI tools have to make our work and lives easier, especially if you are a small business owner with limited financial resources.

Drafting business documents with generative AI tools may be the most cost-efficient, but there are a number of legal pitfalls to look out for as well. This article will cover the following topics:

• What is generative AI?
• Why might a business owner use generative AI?
• What are some business documents that you will need?
• What are the risks of relying on AI to produce business documents?
• What is the best way to prepare business documents?

What is Generative AI?

Generative AI refers to the use of AI to create new content, including text, images, audio, and video. Broadly speaking, generative AI works by training machine learning models to observe and learn patterns in a dataset of human-created content. The models then use the patterns learned to create new content, which, as you would have observed above, is sometimes almost indistinguishable from human-created content.

Examples of the more well-known generative AI tools in the market today include:

• ChatGPT: ChatGPT is a chatbot developed by OpenAI which allows users to “have a conversation” with it, in a way that simulates and mimics a real conversation. Users ask questions or make requests, which serves as a prompt, and ChatGPT will generate responses to and based on that prompt.
• Dall-E: Not to be confused with the cleaning robot from the eponymous film, Dall-E is another AI system created by OpenAI, which is trained to generate and produce images from text descriptions. Users provide short phrases, which serve as the prompt in this case, and Dall-E uses that to create images representing the prompt.
• Bard: Bard is commonly referred to as Google’s answer to ChatGPT, in the sense that it is also a chatbot which generates human-like responses to prompts and queries provided by users.

One of the advantages of generative AI is that there are virtually endless possibilities in terms of its use cases, especially as generative AI tools are increasingly and exponentially becoming more advanced. However, some of the more common applications of generative AI tools presently are:

• Summarising and explaining complex concepts simply.
• Generating content needed for social media or for other modes of marketing.
• Conducting basic research or directing users to relevant research resources.

Why Might a Business Owner Use Generative AI?

As a business owner, you may be interested to explore how generative AI can be a part of your processes, because generative AI tools have the potential to offer unparalleled efficiency and convenience. By implementing generative AI tools in a way that complements your current workflows, you may be able to automate repetitive tasks, hence reducing time and costs spent on such tasks.

For example, you may be able to leverage generative AI to:

• Generate personalised responses to customers and improve customer experience overall.
• Develop new products or brainstorm new ideas for new products based on an analysis of current trends.
• Supplement your existing human resources management, including drafting an internal code of conduct or policy that is tailored to your business needs.

Companies have also started to integrate generative AI tools into their processes. For example, Expedia has a ChatGPT-powered travel planner that lets users ask questions and get recommendations on travel, lodging and activities. Shopify has also added ChatGPT to its platform, which allows store owners to utilise the tool to write content for their social media pages, draft emails to respond to consumers’ queries, and even to generate product descriptions based on what would be most likely to entice consumers.

What are Some Business Documents That You Will Need?

As a business owner, you need to have the right business documents on hand. Even though business documents are usually only used for internal purposes within your organisation, they are still important, as they can affect how your business operates. Some common examples of business documents are:

• Company policies: Company policies include documents like employee handbooks, which set out acceptable behaviour for employees in the workplace, or work from home policies.
• Company bylaws: Company bylaws refers to a legal document setting out the rules and regulations that govern how a company is run. It will also set out basic information about the company, a statement of the company’s purpose, as well as information about the company’s shareholders, directors, and other officers.
• Employment agreements: Employment agreements are contracts between the employer and their employees that define the responsibilities of the employer and employee. They usually also include crucial information such as the length of the employee’s employment with the company, the employee’s contractual working hours, leave or pay entitlement, provisions for the termination of the agreement and severance pay (if any).
• Confidentiality agreements: Confidentiality agreements refer to agreements whereby a party agrees to keep confidential certain information that the party obtains. This is important to protect trade secrets or sensitive data of the company that employees may acquire in the course of their employment. For example, employees of a tech company may have access to sensitive and proprietary information pertaining to the products of the company and would have to sign a confidentiality or non-disclosure agreement to prevent them from being able to share such information freely even after they leave the company.

What are the Risks of Relying on AI to Produce Business Documents?

You can choose to leverage text-to-text Generative AI tools (i.e. a tool that lets you key in a text prompt and generates a text output, e.g. ChatGPT or Bard) to produce the above business documents or any other that your company needs.

However, there are a few risks of relying on generative AI to produce your company’s business documents that you have to be aware of so that you can mitigate or eliminate the risks where applicable:

1. Risk of errors:
   • As generative AI tools are trained on a pre-determined and pre-existing dataset, there is a potential for the data to be outdated, inaccurate and even potentially misleading. In a famous case in the US, two lawyers filed certain papers to court, based on research which they had done using ChatGPT. However, unbeknownst to them, ChatGPT had invented 6 fictitious cases, which the lawyers included in their filings without verifying first. The lawyers and their law firm were fined by the court. As another example, Google’s new AI-powered search tool, Search Generative Experience (SGE) was also discovered to have provided inaccurate public health information. Even though the air quality in major US cities had been at dangerous levels, with certain cities issuing alerts asking residents to limit outdoor activity, when SGE was asked about the air quality index, SGE reported that the air quality was good, citing outdated data.
   • Similarly, the accuracy and utility of the business documents generated using generative AI tools will only be as good as the dataset that the tool is trained on. Hence, you should fact-check that the information that is included in the generated business documents is factually accurate. Apart from this, you should also check the business documents generated against the relevant rules and regulations, to ensure that they are compliant. For example, if you use a generative AI tool to create employment agreements, you should check that the terms in the agreements comply with the regulatory requirements, e.g. According to Singapore’s employment laws, employees’ contractual working hours cannot exceed 8 hours a day, and employees are entitled to 11 paid public holidays a year etc.
2. Risk of data breaches and breaching privacy laws:
   • If, in the process of generating business documents, you input your customers’ personal or sensitive information into the generative AI tool, this could fall foul of data privacy and compliance regulations in Singapore, as generative AI tools are prone to data leaks. For example, it has been reported that ChatGPT was the target of a data breach, which allowed users to see the chat history of other active users.
   • Under the Personal Data Protection Act, businesses and other organisations are required to comply with the various data protection obligations if they undertake activities relating to the collection, use or disclosure of personal data. Such obligations include the obligation to put in place reasonable security arrangements to protect the personal data in your business’ possession, so as to prevent unauthorised access, collection, use and disclosure of such data. Hence, by potentially exposing your customers’ personal data to data breaches through the use of generative AI tools, you may be liable for breaches of data protection laws for which you may be subject to financial penalties.
   • Moreover, generative AI platforms generally make little to no security assurances in their terms of use. For example, OpenAI’s terms of use contain a general disclaimer stating that OpenAI does not warrant that any content (this includes data that you input to ChatGPT) will be secure or not lost or altered. Thus, it is all the more important that you are careful with how you use such generative AI platforms, given that the consequences of any data breaches are likely to fall on your shoulders.
3. Risk of unintentional disclosure:
   • Closely related to the point above, since generative AI relies on building its dataset by accumulating data, even from inputs by other users or your previous inputs, there is a risk that information that you provide may be disclosed in responses to other users.
   • Thus, you may want to avoid inputting commercially sensitive data or information that could inadvertently jeopardise your business, e.g. trade or commercial secrets.

What is the Best Way to Prepare Business Documents?

Despite the potential risks involved with using generative AI tools, they can be a capable and convenient starting point in your preparation of business documents needed for your business. For example, you can use a generative AI tool to create a template for specific business documents, which you can then build on and fill in independently.

You can also use the generative AI tool to get a sense of typical clauses that are included in a specific business document, which you can either adopt wholesale or tweak to suit your business needs. From there, it would ultimately be beneficial to seek legal advice or guidance before you finalise the relevant business documents. A lawyer would be able to help you with the following, among other things:

• Advising you on what business documents you need for your business;
• Reviewing the accuracy of templates or business documents created by generative AI tools;
• Drafting any additional business documents that have to be tailored to your business needs; and
• Ensuring that your business documents are compliant with the applicable rules and regulations, and ensuring that they are regularly updated to account for changes in the law.

—

One of the most compelling advantages of generative AI is efficiency – and you will typically be able to get results faster than you would if you had to do the same task manually. However, when it comes to the preparation of business documents, which can have serious legal implications for your business, you should be aware of the risks of relying entirely on such tools to produce your documents.

In fact, the best practice, and also the best way to reap the benefits of generative AI while avoiding the potholes is to seek advice from experienced commercial and/or contract lawyers, who can assist you as described above. In the event that you are exposed to legal liability arising out of your business documents, the lawyers will also be able to advise you on how best to handle the matter.