The Copyright Crisis in the Age of AI: Pragmatism, Peril, and Solution

[Recommend my two-volume book for more reading]: BIT & COIN:  Merging Digitality and Physicality

The Problem

The rapid ascent of Artificial Intelligence has instigated a profound crisis at the very heart of copyright law. Copyright finds itself fundamentally challenged by technologies designed to ingest and synthesize the entirety of human knowledge, including vast troves of copyrighted material. This clash pits established legal principles against the perceived imperatives of technological progress and global competition, creating a complex dilemma with significant legal, economic, and cultural implications.

Traditionally, copyright infringement hinges on the unauthorized reproduction or adaptation of a protected expression. If a human were to systematically copy, compile, and synthesize copyrighted works on the scale that AI models do during training, it would certainly constitute a massive infringement.

However, AI introduces a significant complication. AI, particularly large language models (LLMs), is trained on vast datasets, often including copyrighted material like books, articles, and codes. Copyright law is rooted in the protection of original, creative human expression, but AI synthesizes this information, not repeating exact expressions, making infringement hard to prove. Recent lawsuits, such as the New York Times suing OpenAI for allegedly reproducing content verbatim (US newspapers sue OpenAI), show this is a growing issue.

LLMs don’t typically regurgitate exact copies of their training data; instead, they identify patterns, relationships, and grammatical and stylistic elements across billions of data points, including copyrighted texts and images. They then synthesize new outputs based on the patterns learned. This process often obscures the direct lineage to any single copyrighted source, making it difficult, if not impossible, to prove substantial similarity under existing legal frameworks, especially given the simultaneous use of countless sources.

(Note: AI “synthesis” isn’t merely mixing information as a person would do manually. It uses a deep neural network architecture to tokenize input information and embed the tokens into a multidimensional vector space to train a transformer function. It then infers output information based on the trained transformer function. It is therefore even more difficult to identify AI’s copyright infringement than doing the same for human derivative works.)

Does this synthesis, built upon the foundation of potentially infringing ingestion, constitute a violation of the copyright law? The law, designed for human-scale creation and copying, struggles to provide a clear answer.

Courts, like in Thomson Reuters v. ROSS Intelligence (2025), are rejecting fair use defenses, suggesting AI’s use is legally problematic. The reasoning of the court seems to be commonsensical.

However, the legal struggles aren’t solving the problem; they are only creating a crisis that needs resolution.

Pragmatic Arguments for AI

This legal ambiguity, however, is increasingly being overshadowed by potent pragmatic arguments. AI is widely perceived as a transformative technology, crucial for economic growth, scientific advancement, and national security.

The argument, particularly potent in geopolitical contexts like the rivalry between the US and China, is that any nation significantly restricting AI’s access to data—even copyrighted data—risks falling behind. The reason is that because copyrighted materials are available to anyone in any nation, one nation’s unilateral decision to stop AI’s free use of materials only puts that nation into a disadvantageous position.

The mantra becomes “AI is too important to be stopped or even slowed.”

From this perspective, the question isn’t whether AI will be allowed to train on existing data, but how the legal system will adapt to accommodate this perceived necessity.

The inclination is to modify or reinterpret copyright law to permit AI’s current practices, prioritizing technological dominance over established creator rights.

Pragmatism, fueled by economic aspirations and fear of competition, appears poised to bulldoze legal tradition.

The reality today is that governments, including the US, are pressured to provide solutions in favor of AI quickly. What the Trump administration has been doing recently (year 2025) strongly indicates a wholesale cave-in.

A Fundamental Flaw of The Pragmatic Argument

Yet, this pragmatic rush ignores critical underlying truths and potential long-term consequences. The core assumption driving the “inevitability” argument – that there’s no practical way to allow AI access while respecting copyright – is flawed.

The real bottleneck isn’t a fundamental incompatibility, but rather the lack of an efficient, scalable mechanism for licensing and compensating creators for the use of their data in AI training.

Existing licensing models, designed for individual transactions or limited scopes, are way too slow, cumbersome, and expensive to handle the petabyte-scale data ingestion required by modern AI. Attempting to negotiate permissions for billions of individual works through traditional means would indeed be too difficult, not to even mention the much harder and much more expensive execution and enforcement. Requiring such would indeed cripple AI development.

But technological solutions do exist. They are just buried behind the noise, and blocked by ignorance. See below section “The Solution”.

The Synthetic Nature of AI and Harm of The Current Path

Furthermore, the prevailing discourse often obscures and overlooks a fundamental characteristic of current AI: it is primarily synthetic, not truly generative or creative in the human sense.

AI excels at recombining, remixing, and statistically optimizing patterns derived from existing human creations. Its impressive power lies in its efficiency at processing and synthesizing the vast accumulation of past human ingenuity. If we reshape our copyright system to allow AI models to freely ingest and monetize derivatives of human creativity without adequate compensation or control for the original creators, we risk profoundly discouraging the very source of future innovation.

Why strive to create original works if they will be instantly absorbed and repurposed by AI systems, diminishing their market value and recognition?

AI is efficient and can effectively generate 99% of human information without losing coherency (AI is Not Generative But Synthetic). This efficiency risks discouraging human creativity, the 1% inspiration vital for innovation, eventually leading to a closed information system at entropic equilibrium (see High-Entropy System Versus Low-Entropy System).

This could lead to an increasingly closed information system, dominated by AI-synthesized derivations of past work. The system would still generate seemingly new information, but the “newness” of the information is just a reflection of the vast information space that is already created by human creativity. New permutations will certainly appear new, and even functionally new, but not new from a fundamental information point of view, as everything comes from a closed system. It will eventually reach an entropic equilibrium (see, High-Entropy System Versus Low-Entropy System), a vast echo chamber endlessly reflecting diminishing returns on human creativity.

But diminishing returns to human creativity isn’t the worst problem. The destruction of the very meaning of human existence is. The very purpose of human existence is to be the universe’s truth finder, the truth tester, and the truth exerciser. Humanity cannot be a closed system, even if such a system can be productive in a materialistic sense. It must be inspired and motivated to receive revelations and make discoveries, to test them, prove them, and act upon them. Without such, humanity would lose its existential meaning.

Therefore, while the legal questions surrounding AI and copyright demand urgent attention, focusing solely on legal adjustments without addressing the underlying technological and economic realities and philosophical implications is inadequate. A robust legal framework that balances the needs of AI development with the rights of creators is essential.

However, even such a framework will still remain vulnerable to pragmatic overrides unless complemented by a capable technological solution that makes fair compensation feasible by enabling automated right management and micropayments at scale.

The Solution

The solution is to shift to the paradigm of decentralized data management leveraging scalable blockchains and smart contracts. Such systems could automate the process of tracking data provenance, verifying usage rights, and executing micro-payments to creators whenever their work is accessed or utilized by an AI algorithm.

Existing payment methods for licensing are slow and expensive, which is unfit for AI’s scale. A suitable decentralized data paradigm using blockchain and smart contracts must be able to automate royalties at large scales, ensuring creators are compensated while AI thrives.

The development and adoption of automated, transparent, and scalable licensing mechanisms, potentially built on decentralized technologies, are not merely technical niceties; they are prerequisites for navigating the AI revolution in a way that fosters continued human creativity and prevents our information landscape from collapsing into an entropic equilibrium state of synthetic repetition. Only by integrating competent technological solutions with legal reform can we hope to resolve the copyright crisis in a manner that serves both innovation and the enduring value of human expression.

The key is automation and low cost, which the traditional systems do not offer. The blockchain solution offers an efficient and low-cost pathway to fairly compensate creators at scale without imposing prohibitive friction on AI development.

Overcoming the Obstacles

The primary obstacles to adopting such solutions are not technological impossibility, but rather widespread ignorance of their potential and the negative connotations associated with the speculative excesses of “cancerous crypto,” which have unfortunately overshadowed the genuinely innovative blockchain for automated rights management and transparent data governance.

But those who have seen the vision do not give up. We have been trying to show the world why a decentralized data paradigm built upon the true Bitcoin blockchain (BSV, not BTC) designed and developed according to Satoshi’s vision is the only viable economic and tech solution to save digital humanity from corruption. The blockchain needed by the new decentralized data paradigm is already here: unbounded scalability, near-zero transaction cost, and full smart contracts, enabling fully automated decentralized identity management, rights management, micropayments, and Universal Single Source of Truth (USSoT). It will be the integrated truth layer of the New Internet based on IPv6. For this purpose, we are not reducing but increasing our efforts in the future.

[Recommend my two-volume book for more reading]: BIT & COIN:  Merging Digitality and Physicality