in OpinionMEMBER EXCLUSIVE
How does AI affect free speech?(Image credit: Westend61 via Getty Images)
- Facebook
- X
- Reddit
- Pinterest
- Flipboard
Share this article 0Join the conversationFollow usAdd us as a preferred source on GoogleSubscribe to our newsletter
Who bears accountability for AI’s output? Technology firms like OpenAI assert they do not. Indeed, their terms of service in 2023 stipulated that accountability rests solely with the end-user. A German court, however, holds a different view.
On June 9, a Munich court, pending potential appeal, determined that Google could be held responsible for erroneous assertions generated by its AI summaries. This ruling establishes a clear distinction between typical search results and statements fabricated by machines. In essence, AI enterprises must face legal consequences for the content produced by their systems and disseminated to users.
The court’s rationale was straightforward yet significant: Search results direct users to external sources, whereas AI summaries articulate statements in Google’s own voice. This divergence is crucial because it addresses the fundamental question of which types of expression warrant protection and which are subject to legal examination. The United States should emulate the German court’s precedent. Without such frameworks, the entire onus of distinguishing factual information from misinformation falls upon the reader.
Akhil Bhardwaj
Akhil Bhardwaj investigates extreme occurrences, encompassing everything from catastrophic organizational failures to groundbreaking innovations. Akhil’s focus lies in the epistemological challenge of comprehending the underlying dynamics that precipitate these events. He also examines methods for enhancing cognitive processes and the ramifications of AI integration within strategic management, entrepreneurship, and high-stakes systems.
In the U.S., the First Amendment is designed to safeguard the liberty to express, debate, persuade, and offend. However, freedom of expression is not absolute. It does not permit incitement to criminal acts, threats, or defamation, for instance. Furthermore, if speech results in tangible harm, the originators can be held accountable for those damages. When a corporation positions an artificial response engine between its users and the internet, it transitions from merely hosting discourse to generating a fusion of intricate mathematical computations that, when presented as text, mimic human communication. AI companies seek the same protections for this text as user-generated content, while simultaneously evading all associated responsibilities of being a speaker.
The origins of this predicament trace back to the 1990s, with the emergence of online forums and social media presenting a novel challenge. Unlike conventional publishers, forum administrators needed to provide a platform for their users’ opinions without assuming liability for the content posted. This issue was addressed through Section 230 of the Communications Decency Act, enacted in 1996. Section 230 was a bipartisan measure intended to preserve the internet as a venue where ordinary individuals could voice their thoughts (or post messages) without the forum operator incurring liability for every message posted by a third party.
This broad immunity reflected a societal consensus: holding platforms accountable for all user-generated content would likely compel many to engage in aggressive censorship or cease hosting speech altogether, thereby curtailing freedom of expression. Section 230 was designed to protect the broader landscape of human expression. In this context, operators of online spaces can be viewed as facilitating a public forum where dialogue takes place.
Freedom of speech is a fundamental human right—it protects individuals as both speakers and listeners within a democratic public sphere.
The legislators who enacted Section 230 three decades ago could not have anticipated a landscape populated by text generated by chatbots. As such content increasingly leads to real-world detrimental effects, litigation is escalating, and tech corporations are employing a variety of often-conflicting legal tactics to evade culpability. In certain instances, they contend that AI-generated text is not speech but merely a tool, thus positioning the companies as protected “carriers” rather than “publishers” under Section 230’s safeguard for a public forum for free expression.
However, the companies resort to this argument only when it serves their interests.
In other situations, they increasingly invoke free-speech principles to defend AI-generated text, as free-speech protections offer extensive legal immunity. For example, in a wrongful-death lawsuit filed in Florida against OpenAI (the developer of ChatGPT), a complainant alleged that the company’s chatbot prompted a 14-year-old to end his own life. OpenAI asserted that the chatbot was shielded by the First Amendment; however, the judge rejected this defense and permitted the case to proceed.
Neither of these contentions is persuasive. AI companies are not simply providers of a public forum; the words generated by their AI summaries and chatbots originate from the company’s products.
Similarly questionable is the assertion that bots should be regarded as equal participants in a public square. This represents a categorization error. Freedom of speech is a human entitlement—it safeguards individuals as speakers and listeners within a democratic public sphere. Bots do not vote, engage in deliberation, voice dissent, practice faith, or participate in civic life. They produce text, but they lack moral and political standing. Bots have no personal stake in the outcomes.
What, then, justifies constitutional protection in the first place? Extending the most robust speech protections to machines would not serve liberty; it would conflate “bot output” with authentic expression. It would, in effect, grant paramount free-speech protection to corporations. However, this necessitates a distinct line of reasoning that society should collectively establish.
OpenAI, the creator of ChatGPT, contended that the chatbot possesses First Amendment protections.
(Image credit: Jaque Silva/NurPhoto via Getty Images)
The Munich court’s measured and discerning approach to regulating “bot output” offers a clear path forward.
Considering its historical context with Nazism, Germany does not uphold free speech with the same fervor as the U.S. Nevertheless, the German court’s arguments provide a valuable framework for a future U.S. ruling.
The Munich court ruled that if a system merely directs users to information sources, it resembles conventional search engines and should retain the extensive protections afforded to aggregators. However, if it synthesizes claims, adopts an authoritative tone, and presents a singular, AI-generated authoritative answer, it should be subject to commensurate duties of care, including corporate liability.
The necessity for such safeguards is escalating. AI-generated summaries can be disseminated instantaneously, scaled globally, and replicated across various platforms until a falsehood is perceived as “truth.” This is not a hypothetical concern; it is already a reality.
Related stories
AI chatbots are turbocharging violence against women and girls: We urgently need to regulate them
AI chatbots oversimplify scientific studies and gloss over critical details — the newest models are especially guilty
AI for breakup texts? How ‘sycophantic’ chatbots are messing with our ability to handle difficult social situations.
Furthermore, it is essential to recall that the original intent of Section 230 was to shield platforms from liability concerning third-party posts, not their own generated content.
This is not an argument against technological advancement. AI can be beneficial, efficient, and genuinely revolutionary. The law should foster the development of useful tools while demanding that the companies deploying them remain accountable for the foreseeable negative consequences of their creations.
We require more precise regulations that maintain an open internet for individuals while preventing machines from legitimizing falsehoods through authoritative presentation. The German ruling indicates such a future. The sooner U.S. legal and policy frameworks align, the greater our prospect of preserving our shared reality and fostering a robust democracy.
Opinion on Live Science provides insights into the most critical scientific issues impacting you and the world today, authored by specialists and leading scientists in their respective fields.
Sourse: www.livescience.com
