Legal Ink: Holding Companies Accountable for AI-Authored Words

Companies wield significant influence through the content they disseminate. Whether it’s a social media post, a blog article, or correspondence with clients, the impact of published material can be far-reaching. However, with great power comes great responsibility, and companies must navigate legal obligations related to the content they put out.

Employer Liability and AI-Generated Content

“Just as newspapers are responsible for their reporters, employers may be liable for content published by their employees”^[i]. The rise of artificial intelligence introduces new complexities as more and more employees are utilizing online tools, often without their employer knowing about it. There is interest in tying Section 230 of the Communications Decency Act of 1996, however,this would only apply to the AI generators themselves, likely not the publishing entity.

Organizations are accountable for the content they publish, whether human-authored or AI-generated. By embracing self-regulation and acknowledging their societal impact, companies can navigate the digital landscape while upholding legal and ethical standards.

Tips for good self-regulation

• Companies should clearly disclose when AI generates content. Users deserve to know if they are interacting with automated systems. Transparency builds trust and allows users to make informed decisions.
  
  For example, the three previous sentences and the article title were both AI-generated.  That said, they have been reviewed and fact-checked by humans, a publishing best-practice.

• Companies should develop comprehensive guidelines for AI content creation and make these guidelines an official part of the employee handbook. Employees should be given clear direction on when and how using AI to assist them with their work is acceptable. These guidelines should address potential biases, harmful language, and misinformation.
  
  We strongly recommend engaging your technology department in this conversation. Because of how AI works, it is very likely that your IT department does NOT want employees feeding company or client-sensitive material into AI generators such as ChatGPT.  There are safer alternatives for how to use AI and all employees should be trained on the organizations’ preferred methods.

• Humans should review all AI content before disseminating.  While AI is generally strong at grammar and can be used as a spelling and grammar check for pre-written content, it is not perfect in its content generation. Hallucinations, where AI creates misinformation, “facts”, and data to prove the point it is trying to make, are still prevalent.  Human review can catch errors, biases, and inappropriate material that AI may miss.
  
  The legal industry is not immune to these problems.  In 2023 a personal injury attorney used ChatGPT to prepare a filing but did not proofread to catch that imaginary cases were being cited in the document. New York-based federal Judge P. Kevin Castel threatened sanctions for the citation of non-existent cases.  The attorney’s excuse? He didn’t realize that the AI generator wasn’t a search engine.

All AI users should understand the legal obligations related to using machine-produced content.  If your organization does not yet have an AI policy, Stall Legal can help you prepare one.

[i] Legal Liability for Published Content. Risk Management Magazine of Risk and Insurance Management Society, Inc.  October 7, 2010 https://natlawreview.com/article/legal-liability-publishedcontent#:~:text=Just%20as%20a%20newspaper%20is%20responsible%20for%20its,and%20broadcasters%20while%20enjoying%20fewer%20privileges%20and%20immunities.