AI Accountability: The Emerging Legal Landscape for Artificial Intelligence

Legal scholars, lawmakers, and even a Supreme Court justice foresee a future where companies are held accountable for the actions and utterances of their artificial intelligence systems, signaling the onset of a wave of lawsuits.

If your organization utilizes AI for content creation, decision-making, or impacting people’s lives, the likelihood is that you’ll bear responsibility for its actions, particularly in instances of error.

This extends to major tech players deploying chat-based AIs to the public, such as Google and Microsoft, alongside well-funded startups like Anthropic and OpenAI.

Jane Bambauer, a law professor at the University of Florida, highlights the inevitability of companies facing some form of liability if the prevalent trend of leaning on AI for content creation and judgment continues.

every entity utilizing generative AI could find itself accountable under laws governing harmful speech and defective products, given that contemporary AIs function as both speech creators and products.

The repercussions are significant: every entity utilizing generative AI could find itself accountable under laws governing harmful speech and defective products, given that contemporary AIs function as both speech creators and products. Legal pundits predict a surge in lawsuits across various business scales.

The consequences of AI-generated output transcend mere damage to a company’s reputation; concerns about future liability prompt companies to tweak their systems to sidestep problematic outputs. For instance, Google’s Gemini faced criticism for appearing overly “woke.” Additionally, efforts to curb “hallucinations,” where generative AIs fabricate content, are on the rise within the industry.

From a legal standpoint, Section 230 of the Communications Decency Act of 1996 historically shielded internet platforms from liability for user-generated content. However, this protection doesn’t extend to content generated by a company’s AI, exposing tech firms to unprecedented legal risks.

Graham Ryan, a litigator at Jones Walker, emphasizes the uncharted legal territory presented by generative AI, contrasting it with previous eras of the internet.

Legal experts unanimously agree that Section 230 won’t shield companies from lawsuits over AI-generated outputs, which now encompass not only text but also images, music, and video.

Furthermore, potential defendants extend beyond major tech players to include companies incorporating generative AI into their services, broadening the scope of responsibility across various sectors.

Supreme Court Justice Neil Gorsuch’s remarks in early 2023 underscore the likelihood that current laws won’t safeguard AI companies or their users from legal accountability.

As AI companies engage in legal debates over issues like scraping copyrighted content, arguments claiming substantial transformation of content may inadvertently weaken assertions of non-responsibility for generated content, potentially eroding the protection of Section 230.

Courts vs. Congress

Traditionally, companies would lobby Congress to address gaps in existing legislation. However, recent shifts indicate a reluctance from Congress to maintain the full protections of Section 230, suggesting that companies must adhere to stricter regulations to benefit from its safeguards. This stance contrasts sharply with the desires of AI developers and users.

Adam Thierer, a senior fellow at the conservative think tank R Street Institute, notes a prevalent sentiment among lawmakers to significantly amend Section 230, presenting a challenging legislative landscape for companies in the AI sector. Consequently, these entities lean towards maintaining the status quo and abstaining from advocating for legislative revisions.

Jane Bambauer, hailing from the University of Florida, proposes a nuanced approach to applying existing laws to AI, advocating for a case-by-case examination in courtrooms. She posits that allowing legal proceedings to unfold incrementally, alongside emerging insights into the impacts of AI, could strike a balance between fostering innovation and curbing potential abuses.

Navigating Liability in the Age of AI

OpenAI finds itself entangled in defamation lawsuits, notably one where a Georgia radio host alleges the company’s chatbot falsely accused him of embezzlement. OpenAI contends that it bears no responsibility for the content its chatbot generates, likening its product to a word processor—a tool for users to create content.

However, Jason Schultz, director of New York University’s Technology Law & Policy Clinic, doubts the viability of OpenAI’s argument. He asserts that while Microsoft Word provides a blank canvas, it doesn’t furnish pre-scripted essays.

As legal experts grapple with the potential harms AI may inflict, it’s evident that speech isn’t the sole liability companies face when employing generative AI. These systems, designed with inherent biases, can inadvertently yield harmful outcomes, from biased hiring practices to dispensing inaccurate advice, leading to financial repercussions for unsuspecting individuals.

Given the multifaceted applications of AI across diverse industries, understanding its potential harms and devising effective regulations will require time, explains Schultz.

However, the prevailing legal uncertainties surrounding companies utilizing generative AI, coupled with ensuing compliance challenges and litigation, pose significant risks, according to Ryan of Jones Walker. This perceived threat, as highlighted by Thierer of R Street Institute, jeopardizes the advancement of AI as a whole.

Alternatively, some suggest that mitigating the legal risks may involve curtailing the use of current generative AI tools.

Alternatively, some suggest that mitigating the legal risks may involve curtailing the use of current generative AI tools. Michael Karanicolas, executive director of the Institute for Technology, Law & Policy at UCLA, posits that if the proliferation of AI chatbots results in a surge of lawsuits, companies may opt to restrict access to such technology.

He argues that imposing costs or liabilities for AI-induced harms, or rendering the technology economically unviable, could serve as a deterrent against misuse and incentivize responsible AI development.