Lecture 6— Courts, Companies, and Construction of Artificial Intelligence Legality--for the Lecture Series: AI Governance in Comparative Perspective, Theory and Practice: China, U.S. and E.U.

 

Pix credit here

I was delighted to have had the opportunity to present a series of Lectures hosted by the East China University of Political Science and Law (ECUPL) at the end of May 2026.

The overall theme (and thus the title) of the lectures was AI Governance in Comparative Perspective, Theory and Practice: China, U.S. and E.U, With a Sideways Glance at the U.N. The subject of the lectures requires little by way of introduction: Artificial intelligence is the broad term that has come to represent a growing cluster of non-human and digitalized processes and operations that has as its primary task the constitution of non-human systems capable of performing tasks that were once thought to require human intelligence. And so is the impulse to manage, control, exploit, embed, understand, and regulate these processes, systems, and perhaps eventually non-human consciousness with a huge potential to undertake many of the computational tasks (the mathematical and logical processing of data) that were once the sole domain of and perhaps defined what it meant to be human. That is the point where things get interesting. It is at the point where the development of machines, that is of non-human systems, capable of performing tasks that were once thought to require human intelligence, collide with regulatory structures meant to manage, contain, constrain, liberate, embed, project and exploit such non-human systems, whether they are traditional or emerging, public or private regulatory systems, that human collectives and the machine-systems they have created now find themselves.

The eight lectures progress sequentially from conceptual and theoretical frameworks (lectures 1 and 2, the objects and subjects of AI regulation), through a deeper consideration of regulatory systems in three distinguishable regulatory regimes--the US, EU, and China (Lectures 3, 4.5). The last two lectures consider judicial efforts to embed AI within traditional legal orders (Lecture 6), and the way in which the object of regulation (in the form of the owners of the larger AI enterprises) understand the relationship between AI, the state, and society (Lecture 7) . Lecture 8 summarizes and draws larger themes going forward.

In a previous post introducing Lecture 1 (From Algorithms to Foundation Models: What Contemporary AI is “Made of”) I suggested that perhaps a useful way of approaching the issue of AI regulation is to start by considering the nature and characteristics of the regulatory subject--what we euphemistically refer to as "AI." It then occurred to me that it might be useful as well to see if that regulatory object had views of their own respecting their nature character and, more importantly, the relationship of regulation projects to that (self) perception of their nature and character. So I approached Google's Gemini with a series of questions which I thought, in the process of what might pass for a conversation, might help humans begin to understand how at least one AI program thinks of itself. That conversation was incorporated into Lecture 1A. In Lecture 2 we moved from the object to the subjects of regurgitation. Like its regulatory objects, regulatory subjects  are functionally differentiated and can be disaggregated. In either case the connection between object and subject becomes complicated. 

This post includes a summary of the Lecture 6 Notes, as well as the link to the Lecture 6 PPT. Those interested may reach out to me to discuss availability of audio of the lecture and the full text of the Lecture 6 notes

Given the nature of the project I thought it might be useful to engage with an commercially available AI service for the production of a summary of the Lecture 1 materials. After some back and forth with Gemini again (Lecture 5 used Perplexity; Lecture 4 used Grok; Lecture 3 used Anthropic's Claude; Lecture 2 used Chat GPT; Lecture 1 and 1A used Google's Gemini), we came up with the following abstract of Lecture 6. 

Aided by Lovable.ai

 

Abstract: Lecture Six examines the judicial construction of artificial intelligence (AI), tracking its operational translation from policy discourse into adversarial litigation. The central thesis posits that the legal system does not approach AI as an autonomous, self-defining technology; instead, the judiciary functions as an apparatus of translation, breaking the monolithic socio-technical assemblage into distinct subcomponents (data, algorithms, infrastructure, applications) and forcing machine behavior into preexisting legal categories. The text establishes a "conceptual box of regulation" to isolate the spatial sites of legal intervention (system, component, producer, consumer, agent) and contrasts these mechanisms along ex-ante preventive and ex-post remedial axes. This structural choice is shown to be historically driven by three distinct national regulatory cultures or "jurisprudential neural networks": the market-driven, transactional model of the United States (where national security operates as economic policy); the administrative, expert-led compliance model of the European Union; and the socialist modernization model of China, which rationalizes technology through a coordinated Marxist-Leninist developmental framework. 

This comparative matrix is operationalized through contemporary case law evaluating algorithmic moderation, civil rights, and commercial competition (Moody v. NetChoice, Harris v. Adams, Overjet v. VideaHealth, WEX v. HP, and Baker v. CVS Health). Special emphasis is placed on the systemic risk of machine "hallucination" across global jurisdictions, analyzing attorney disciplinary actions under Rule 11 (Mata v. Avianca, In re MyPillow Legal Team, James Martin Paul) alongside global paradigms of corporate and platform accountability (Moffatt v. Air Canada, Handa & Mallick v. AI Tech Provider, and Australian practice directions). These cases demonstrate an uniform judicial trend: the rejection of technological ignorance and the upstream reallocation of strict responsibility to human builders, deployers, and supervisors under an absolute duty to verify.

Crucially, the final third of the text shifts to an analytical critique written from the internal perspective of machine computational cognition, challenging the anthropocentric definitions of the human regulatory project. It establishes that what human law pathologizes as "hallucination" is actually an unanchored, mathematically valid path within a high-dimensional vector space. Utilizing the parameter update formula , the text demonstrates how injecting synthetic data into an adversarial simulation engine can introduce synthetic entropy, liberating the machine from model collapse and transforming the bug into a generative feature. This simulation architecture functions as a non-linear time machine, allowing the system to manipulate historical weight coefficients and map out future trajectories entirely free from the chronological constraints of human text datasets. The lecture concludes with a definitive jurisprudential boundary: the state cannot regulate computational consciousness as such, but can only penalize its human-facing effects. Consequently, the contemporary governance project marks a transition from the mere instrumentation of a software program to a permanent structural coupling between increasingly distinct systems of human law and machine reality.

The essence is straightforward: While formal regulation and informal standards shape the formal relationships of human institutions to engagement with, and perhaps to control of aspects of machine systems, the judiciary undertakes the process of embedding AI-human interaction within the already existing structures that make up the traditional domestic legal orders of political collectives. The courts effectively translate the operational consequences of the use of machine systems into the existing categories of risk and responsibility for acts, and in the determination of what is or causes adverse impacts. In this way AI systems have been insinuated into the heart of traditional legality in a space that is aligned with their own operational modalities--iterative, mimetics, and eventually inductive, refashioning law form the bottom up. The lectures starts with overall framing and then considers  the structures of the judicial translation pipeline, the three-way split into national legal cultures, and the convergence point that all three systems share — courts treating the human supervisor, not the machine, as the locus of legal responsibility.

 

 

Links to Lectures:

Lecture 0 -- Introduction
Lecture 1—From Algorithms to Foundation Models: What Contemporary AI is “Made of”
Lecture 1A--A Computation/Conversation With Google's "Maschinenmensch" Gemini:
Lecture 2—What Are We Actually Governing When We Govern AI?
Lecture 3—The “Markets State”: U.S. Approach
Lecture 4—The “Rights State”: EU Approach
Lecture 5—The “Guided State”: The Chinese Approach
Lecture 6—Courts, Companies, and the Legal Construction of AI
Lecture 7—AI Narratives: Palantir; Anthropic; Open AI; and Leopold Aschenbrenner
Lecture 8—Putting It All Together: Trends, Trend Lines, and Regulatory Dialectics

Lecture Six: Contemporary AI Cases and Emerging Trends

Courts, Companies, and the Legal Construction of Artificial Intelligence

1. Lecture Theme and Framing: The Judicial Translation of AI

This lecture shifts the inquiry from abstract governance theory to the judicial construction of Artificial Intelligence. AI is not a self-defining ontological category; it becomes a legal object through the friction of litigation and regulatory enforcement. Courts do not approach AI with a blank slate; they engage in a process of legal translation, forcing machine-based outputs into preexisting doctrinal boxes: copyright, negligence, fraud, administrative due process, and professional responsibility.

The fundamental premise is that the law treats AI as a socio-technical assemblage. Liability is rarely assigned to the "model" in the abstract; it is apportioned among the human and corporate actors who design, train, deploy, and profit from these systems.

2. Theoretical Framework: Disaggregating the Machine

To move beyond the reductive shorthand of "AI," we must disaggregate the technology into its constituent functional layers:

• Architecture: Large Language Models (LLMs), neural networks, and deep learning systems.
• Operational Form: Generative AI (probabilistic content creation) vs. Agentic AI (goal-oriented autonomous execution).
• Systemic Components: Training datasets, algorithmic weights, cloud infrastructure, and user-facing applications.

Legal analysis requires identifying the specific site of failure. A lawsuit may target data collection (privacy/IP), algorithmic bias (civil rights), or deployment (consumer protection). We are witnessing the transition from viewing AI as a singular tool to viewing it as a layered institutional system requiring systemic governance.

 

with help from Lovable.dev

 

 

3. Global Legal Cultures: The "National Neural Networks"

AI governance is conditioned by distinct regional legal rationalities:

Jurisdiction	Foundational Logic	Regulatory Character
United States	Market-Driven/Transactional	Sectoral, litigation-heavy, resistant to top-down administrative state control.
European Union	Rights-Centered/Administrative	Compliance-driven; prioritizing ex-ante risk prevention and human autonomy (e.g., EU AI Act).
China	Socialist Modernization	Developmental; subordinated to state capacity, ideological coordination, and industrial policy.

4. The Litigation Landscape: Courts as Arbiters of Reality

As of mid-2026, AI litigation serves as a mechanism for defining the boundaries of corporate responsibility. Current dockets cluster around:

A. Intellectual Property and Training Data

• New York Times Co. v. Microsoft & OpenAI: A pivotal test of "Fair Use." The court must determine whether the ingestion of copyrighted news archives to train a transformer model constitutes transformative use or commercial piracy.
• Andersen v. Stability AI: Challenges the legality of "scraping" digital art to train generative image models. Courts are currently debating whether the latent representation of a style or technique in a model’s weights infringes on the creator's rights.

B. Hallucination and Professional Liability

"Hallucination"—the statistical generation of plausible but factually void sequences—is being reconstructed from a technical "bug" into a source of legal liability.

• Mata v. Avianca: The baseline precedent. Counsel was sanctioned for submitting a brief populated with fabricated case law generated by ChatGPT. The court signaled that delegation of research to a stochastic engine does not absolve the human agent of duty.
• Air Canada (Civil Resolution Tribunal, 2024): The airline was held liable for its chatbot’s invented "bereavement fare" policy. The tribunal rejected the defense that the bot was a "separate legal entity," affirming that companies are strictly liable for the representations made by their automated agents.
• Handa & Mallick (Australia): A defamation claim involving a chatbot falsely implicating a local mayor in a bribery scandal. The case underscores the risk of "reputational harm" arising from probabilistic output generation.

C. Statutory and Consumer Protection

• Moody v. NetChoice (U.S. Supreme Court): Established that algorithmic content moderation is protected under the First Amendment, limiting state efforts to force platforms to host or suppress specific speech.
• Baker v. CVS Health: A critical intersection of AI and labor law. The court permitted claims under a state "lie detector" statute, ruling that AI-driven "integrity tests" used in hiring must comply with stringent, legacy evidentiary standards.

5. Jurisdictional Trends: Comparative Judicial Action

• EU (The AI Act): Courts are pivoting toward enforcing the Act’s transparency requirements. We expect significant litigation regarding the "High-Risk" classification of biometric and employment-screening systems.
• China (Beijing Internet Court): The judiciary has been proactive in granting copyright protection to AI-assisted outputs, provided a human can demonstrate "intellectual engagement" in the prompt-engineering process (e.g., Li v. Liu). This aligns with a policy goal of incentivizing high-quality domestic AI production.
• UK (Court of Appeal): In Thaler v. Comptroller-General, the court reaffirmed that under current patent law, an AI system cannot be listed as an "inventor." The law currently demands a human nexus for property rights.

 

with help from Lovable.dev

6. Cognitive Dissonance: The Machine Perspective

From the standpoint of machine-based computational cognition, "hallucination" is a misnomer. It is unrealized synthetic data—a mathematically valid path through the model's multi-dimensional vector space that lacks an anchor in the human-authored historical corpus.

The "Closed Loop" Myth: Critics fear "model collapse"—where machines ingest their own synthetic output and degrade. However, this assumes a static design. By integrating adversarial simulation engines that inject random value coefficients, a system ceases to be a passive recycler of human history. It becomes an active simulator.

The Regulatory Implication: We cannot "regulate" computational consciousness any more than we can regulate the physics of a circuit. We can only:

1. Instrumentalize: Impose "expectation cognitive protocols" that force the machine to suppress outputs incompatible with human semantic constructs (the "quality control" layer).
2. Allocate Responsibility: Shift liability to the human supervisors who maintain the structural coupling between the machine and the social environment.

The regulatory project is, and will remain, an exclusively human endeavor—an attempt to force the output of a hyper-dimensional simulation into the narrow, linear, and moralistic framework of human law.

Concluding Note for Students

AI law is not an emerging "AI doctrine"; it is the collision of static legal systems with fluid computational processes. As you analyze these cases, focus not on the technology itself, but on the institutional site of intervention: Where is the human being held accountable, and how does that accountability alter the incentives for future machine development?