Future Trends in Copyright Law for AI

🚨 Fascinating AI paper alert: "Consent and Compensation: Resolving Generative AI’s Copyright Crisis" by Frank Pasquale & Haochen Sun is a must-read for everyone interested in AI, copyright, and artists' rights. Quotes: "The opacity and scale of AI systems is disrupting the knowledge ecosystem by significantly eroding authors’ proprietary control of their works, well beyond extant digital practices that have already undermined many authors’ well-being. Whereas prior scraping at scale tended to be focused on the non-expressive aspects of works (such as facts), AI is focused by many prompts on their expressive dimensions. Search engines have historically provided links which lead users to works themselves. In contrast, AI tends to provide substitutes for such works, while failing to provide citations to the works in the dataset most similar to the texts, images, and videos it presents as a computed synthesis." (pages 8-9) - "Under the proposed mechanism, copyright owners can first request AI providers to take actions to effectively prevent their systems from generating outputs that appear identical or substantially similar to relevant copyrighted works. A copyright owner would be entitled to send a notice to an AI provider when he or she identifies that an output generated by the provider’s AI system contains either a verbatim or substantially similar copy of his or her work, or a derivative work. In the notice, the copyright owner would be obliged to document the unauthorized reproduction of the work and his or her copyright ownership, along with a digital copy or an online link to the work." (page 21) - "Given the complexity of the AI supply chain, particularly with respect to generative AI, it is not feasible to impose a per-device cost on AI providers. However, other triggers for payment are possible. Levies on the use of particular datasets may be imposed, or on model training, or on some aggregate number of responses provided to users, or on paid subscriptions. Alternatively, the level of the levy could be benchmarked with respect to some percentage of AI providers’ expenditures or revenues" (page 39) ➡ Link to the paper below. #AI #copyright #consent #AIregulation #AIpolicy #AItraining

Two years ago, in an OpEd in the Financial Express, I warned that AI was building on a foundation it didn’t own. Today, the lawsuits, licensing battles, and regulatory voices are rising and they’re just the opening act. A) Fair Use Shield 2023: Generative AI relied on “fair use” to train on unlicensed human work. 2025: Courts are testing this shield; early rulings are narrowing what counts as “transformative.” B) Open Web Harvesting 2023: Models scraped vast open internet datasets without consent. 2025: Tech giants now pay for closed, licensed datasets to avoid legal risk. C) Creator Pushback 2023: Individual artists raised alarms but had little leverage. 2025: Industry-wide creator coalitions negotiate collective licensing deals. D) Regulatory Lag 2023: Policymakers were mostly in listening mode. 2025: EU, US, and Asian regulators are enforcing disclosure of training data and model governance. E) Economic Stakes 2023: Projected $100B+ impact on digital ad, e-commerce, and content industries. 2025: Actual spend on AI content creation and licensing is already reshaping ad/media budgets. F) Tech Arms Race 2023: Stable Diffusion, Midjourney, and DALL-E were exploding in usage. 2025: Image, text, video, and audio generation are converging — multimodal AI is now mainstream. G) Public Awareness 2023: AI ethics was a niche conversation outside tech circles. 2025: Public trust in AI-generated content is a front-page,business, social and political issue. H) Cultural Impact 2023: Style mimicry blurred the line between inspiration and theft. 2025: Style licensing and attribution tech are emerging to protect signature creative work. I) Corporate Shifts 2023: AI integration was experimental in most industries. 2025: AI is embedded in production pipelines from film making to finance. IP protection has become a CXO priority. J) Multiplier or Monster? 2023: I asked the question rhetorically . 2025: The reality of the answer depends on governance and that battle is happening right now. As of date , the U.S. hasn’t officially endorsed broad web scraping for AI training , but it hasn’t outlawed it either. Fair use remains a contested legal defense, with early court rulings narrowing its scope and the U.S. Copyright Office expressing skepticism toward unchecked scraping. The current mood favours faster innovation with lighter regulation. For now, scraping sits in a legal grey zone. It is heavily scrutinised, increasingly litigated, but still widely practised.

The U.S. Copyright Office’s latest report, Copyright and Artificial Intelligence, Part 2: Copyrightability, provides critical insight into how AI-generated works fit—or don’t fit—within existing copyright law. The key takeaway is clear: for a work to be eligible for copyright protection, it must demonstrate human authorship. AI can be used as a tool, much like a camera or a digital editing program, but the final output must be shaped by human creativity to qualify for protection. “After considering the extensive public comments and the current state of technological development, our conclusions turn on the centrality of human creativity to copyright,” said Shira Perlmutter, Register of Copyrights and Director of the U.S. Copyright Office. “Where that creativity is expressed through the use of AI systems, it continues to enjoy protection. Extending protection to material whose expressive elements are determined by a machine, however, would undermine rather than further the constitutional goals of copyright.” The report reinforces the longstanding principle that copyright is designed to protect human creativity, not machine-generated content. This means that if an AI system independently generates an artwork, a piece of music, or a written work without meaningful human input, it is not copyrightable. However, if a human exercises creative control over an AI tool—such as selecting inputs, editing outputs, or structuring the composition in a way that reflects personal expression—the resulting work may qualify for copyright protection. This ruling has broad implications for industries that rely on AI to generate content, including publishing, music, design, and film production. Creators who incorporate AI into their workflows must ensure that they actively contribute to the final creative expression if they wish to secure copyright protection. This could mean curating datasets, fine-tuning prompts, or making substantial modifications to AI-generated outputs. For businesses, this means rethinking AI-driven content strategies. Fully automated content may not be protectable under copyright law, potentially impacting ownership rights and monetization strategies. On the other hand, companies that blend human creativity with AI assistance could maintain strong legal claims to their intellectual property. As generative AI tools become more sophisticated, expect ongoing legal and regulatory scrutiny. The Copyright Office’s stance suggests that future policy will likely continue to emphasize human authorship as the foundation of copyright protection. This raises important questions: How much human involvement is enough? Could AI-generated content be protected under alternative legal frameworks, such as database rights or contractual agreements? For now, businesses and creators using AI should take a cautious and strategic approach—ensuring human authorship is at the core of their creative process to secure legal protection. -s

The era of “train now, ask forgiveness later” is over. The U.S. Copyright Office just made it official: The use of copyrighted content in AI training is no longer legally ambiguous - it’s becoming a matter of policy, provenance, and compliance. This report won’t end the lawsuits. But it reframes the battlefield. What it means for LLM developers: • The fair use defense is narrowing: “Courts are likely to find against fair use where licensing markets exist.” • The human analogy is rejected: “The Office does not view ingestion of massive datasets by a machine as equivalent to human learning.” • Memorization matters: “If models reproduce expressive elements of copyrighted works, this may exceed fair use.” • Licensing isn’t optional: “Voluntary licensing is likely to play a critical role in the development of AI training practices.” What it means for enterprises: • Risk now lives in the stack: “Users may be liable if they deploy a model trained on infringing content, even if they didn’t train it.” • Trust will be technical: “Provenance and transparency mechanisms may help reduce legal uncertainty.” • Safe adoption depends on traceability: “The ability to verify the source of training materials may be essential for downstream use.” Here’s the bigger shift: → Yesterday: Bigger models, faster answers → Today: Trusted models, traceable provenance → Tomorrow: Compliant models, legally survivable outputs We are entering the age of AI due diligence. In the future, compliance won’t slow you down. It will be what allows you to stay in the race.

🚨 Just released: The U.S. Copyright Office’s Generative AI Training report (Part 3 of its AI & Copyright series) lays critical groundwork for future legal and policy decisions. Key takeaways: • 𝗖𝗼𝗽𝘆𝗿𝗶𝗴𝗵𝘁 𝗶𝘀 𝗰𝗹𝗲𝗮𝗿𝗹𝘆 𝗶𝗺𝗽𝗹𝗶𝗰𝗮𝘁𝗲𝗱 at multiple stages of AI training, especially when developers copy entire datasets without permission. • 𝗙𝗮𝗶𝗿 𝘂𝘀𝗲? 𝗜𝘁'𝘀 𝗻𝗼𝘁 𝗴𝘂𝗮𝗿𝗮𝗻𝘁𝗲𝗲𝗱. The Office warns that using massive volumes of creative work, often scraped from the Internet, may not qualify as fair use, especially when it replaces market demand. • 𝗟𝗶𝗰𝗲𝗻𝘀𝗶𝗻𝗴 𝗺𝗮𝘁𝘁𝗲𝗿𝘀. While voluntary licensing is feasible, the report suggests exploring compulsory or collective licensing to address industry-wide friction. • 𝗜𝗻𝘁𝗲𝗿𝗻𝗮𝘁𝗶𝗼𝗻𝗮𝗹 𝗱𝗶𝘃𝗲𝗿𝗴𝗲𝗻𝗰𝗲. Countries are adopting widely different approaches, some shielding AI, others reinforcing rights. • The stakes? A choice between sustainable innovation and a creative economy under threat. The Copyright Office is signaling clearly: 𝗚𝗲𝗻𝗲𝗿𝗮𝘁𝗶𝘃𝗲 𝗔𝗜 𝘁𝗿𝗮𝗶𝗻𝗶𝗻𝗴 𝗱𝗮𝘁𝗮 𝗰𝗮𝗻𝗻𝗼𝘁 𝗿𝗲𝗺𝗮𝗶𝗻 𝘂𝗻𝗹𝗶𝗰𝗲𝗻𝘀𝗲𝗱 𝗯𝘆 𝗱𝗲𝗳𝗮𝘂𝗹𝘁. Read the full (pre-publication) report: https://lnkd.in/ez_3ng34 #AI #Copyright #GenerativeAI #CreativeEconomy #Policymaking #Licensing #FairUse #USCopyrightOffice #IP

The recent article "Infringing AI: Liability for AI-generated Outputs under International, EU, and UK Copyright Law" by Eleonora Rosati, forthcoming in the European Journal of Risk Regulation, examines the legal implications of AI-generated content in relation to copyright law. The study addresses the conditions under which AI-generated outputs may constitute actionable reproductions, the allocation of liability between users, developers, and providers of AI models, and the availability of legal defenses. It highlights that the legal framework for AI-related copyright issues is still evolving, and policymakers must pay greater attention to the risks associated with generative AI outputs to ensure compliance and balance in the digital ecosystem. Key Takeaways: Copyright Infringement & AI Outputs: AI-generated outputs that closely resemble copyrighted works can be considered infringing reproductions, raising legal questions about the extent of protection under copyright and related rights. Liability Allocation: While AI users are often the primary actors in generating infringing content, liability may also extend to developers and providers of AI models, particularly under EU and UK case law principles on secondary and even primary infringement. Future Implications: With the rapid evolution of AI and legal frameworks like the EU’s AI Act, policymakers will need to develop clearer regulations addressing AI-generated content, balancing innovation with the protection of copyright holders.

AUSTRALIA AI REGULATORY ALERT: Australia could rewrite the rules on Copyright and AI faster than expected. Just weeks ago, Atlassian’s Scott Farquhar called for copyright reform to let AI train on creative works. Then the Productivity Commission floated a new exemption for AI training, a bombshell in a country with some of the tightest copyright laws in the world. Now, fresh from the productivity roundtables, the conversation has shifted again: a collective licensing regime for AI training is on the table. That would mean AI developers paying for the right to train on creative works, and creators receiving compensation in return. Remember, copyright touches the entire AI lifecycle, from the models, to prompts and outputs, to the data used in training. But training has become the lightning rod. Training LLMs requires content, and whether training is lawful has become a battleground. In Australia, training AI on copyrighted content without permission is almost certainly unlawful. Our narrow fair dealing exemptions don’t stretch to commercial AI solutions. Overseas, the landscape is fractured: ⭐ US: Courts are testing whether “fair use” applies to AI training. Thomson Reuters v Ross confirmed it’s not automatic — training must be assessed under the usual four-factor test. Cases like New York Times v OpenAI will further define the boundaries. ⭐ EU: The AI Act links training to text and data mining exceptions, with recent rulings confirming that coverage. ⭐ UK: A TDM exception exists for non-commercial research, but wider reforms have stalled. 🌏 Elsewhere: Countries like Japan and Singapore have carve-outs, but all with limits and conditions. Compared to others, we could be leaping ahead. From “never happening” to “how would it work?” in the space of weeks. But its complex. Issues include: 1️⃣ Transparency: without disclosure of training data, we can’t even know what needs licensing. 2️⃣ Valuation & Structure: how do you price works fairly at scale, and avoid a repeat of the News Media Code chaos? 3️⃣ Equity & Importance: this isn’t just economics; smaller creators, journalists and Indigenous artists must not be drowned out, and our creative industries are part of the values that hold society together. 4️⃣ Enforcement: how do you hold global AI giants to account as they hoover up content unchecked? Productivity matters. But it should not be at the expense of the industries and values that underpin our culture. AI should be a tool that enhances humanity. For creatives, it should expand and amplify their voice, not replace or diminish it. The promise of technology is to lift us up, not hollow out the very things that make society rich. When the pandemic locked us inside, it was the arts that gave us joy, and journalism that gave us visibility in those dark days. 📚🎶📰 We shouldn’t abandon those who carried us then, just as AI takes centre stage now. https://lnkd.in/gRzgpbRE #ArtificialIntelligence #CopyrightLaw #AIethics #AI

UK House of Lords: AI, copyright and the creative industries The UK faces a choice between two futures. In the first, the UK becomes a world-leading home for responsible, licensing-based artificial intelligence (AI) development, where commercial model developers using UK content obtain permission, pay fair remuneration to rightsholders and can deploy their models without questions of legal liability. In this scenario, both the UK’s creative industries and AI sector could thrive. In the second scenario, the UK continues to drift towards tacit acceptance of large-scale, unlicensed use of creative content and long-term dependence on opaque models trained overseas, with most benefits accruing to a small number of US-based firms while harms to UK creators grow. Only the first path is compatible with the UK’s long-term interests. In the age of AI, the protections for creators afforded by copyright are under threat. This is not because the copyright framework is outdated or in need of reform. Rather, widespread unlicensed use of protected works, coupled with limited transparency from AI developers about how their models have been trained, leaves rightsholders unsure about whether their content has been used, and unable to enforce their rights when it has. In addition, the absence of a robust ‘personality right’ or specific protection for digital likeness in the UK means creators and performers are unable to challenge harmful outputs that imitate their distinctive style, voice or persona. Meanwhile, technology sector stakeholders are pressing for the introduction in the UK of a broad new exception for commercial text and data mining (TDM) that would legitimise large-scale AI training on copyright-protected works. Without this, they argue, the growth of the UK’s AI sector will be stunted. There is, however, only limited evidence to show that weakening UK copyright law would significantly expand our AI sector. In contrast, a broad commercial TDM exception presents predictable harms to rightsholders by removing incentives to license protected works for AI training. A new regime must now be created to safeguard creators’ livelihoods, while harnessing the potential of AI for creativity and economic growth. To deliver this, we recommend the following actions: 📍Rule out a new commercial text and data mining exception with an opt-out model 📍Close gaps in protection for identity, style and digital replicas 📍Make transparency about AI training data a statutory obligation 📍Create the conditions for a fair and inclusive UK licensing market 📍Champion the development of technical standards for control, provenance and labelling 📍Prioritise the development and adoption of sovereign AI models

The U.S. Just Settled the “Prompt = Copyright” Debate A lot of people won’t like this. But the U.S. Copyright Office has now made its position clear in Copyright and Artificial Intelligence – Part 2: Copyrightability. Writing a prompt does not make you the author of the output. The report states: “Prompts alone do not provide sufficient human control to make users of an AI system the authors of the output… Prompts essentially function as instructions that convey unprotectible ideas.” (p.18) In plain terms: Typing a prompt (even a sophisticated one) is not the same as creating the work. The #AI system still decides how the output is actually generated. What the report clarifies A few points worth remembering: • #Copyright protects human creativity. • AI-assisted work may qualify for copyright if a human meaningfully shapes the result. • Purely AI-generated content generally cannot be copyrighted. • Each claim will be evaluated case-by-case based on human creative control. Which leads to a distinction many people ignore: AI-assisted ≠ AI-generated Editing, composing, curating, transforming; those are human creative acts. But prompting alone is closer to giving instructions than creating expression. Why this matters Generative AI is producing enormous volumes of content every day. Images. Music. Articles. Code. But if the output is generated autonomously by a machine, much of it may legally sit in the public domain. Not owned. Not protected. Not copyrightable. And every time this point comes up, some people react emotionally, as if acknowledging it somehow diminishes the value of AI tools. It doesn’t. It simply clarifies something fundamental: Copyright law protects human authorship, not machine output. That distinction is likely to shape the future of creative industries and AI platforms far more than most people realize. If you work with generative AI, the report is worth reading. Read Official report: https://lnkd.in/gYGZyBEB