A Creator Updates Their License to Prohibit All AI Derivatives of Their Work

The Anxiety Behind a Licensing Statement

An artist recently updated the licensing terms on their personal website, explicitly prohibiting any form of AI-derived use across all of their artwork, worldbuilding lore, stories, and comics. After being submitted to Hacker News, this seemingly brief declaration drew attention — reflecting the widespread concerns that creators today hold toward generative AI.

Because the original material is limited in detail (essentially a headline-level statement with 8 upvotes and 1 comment), this article focuses more on the broader industry context rather than a deep dive into this single declaration.

Why Creators Are Blocking AI in Their Licenses

As text-to-image and text-to-text models have rapidly improved, the sourcing of training data has become a flashpoint. Vast amounts of publicly available illustrations, fiction, and comics have been incorporated into training datasets without authorization, leaving original creators feeling their labor has been appropriated without compensation.

Explicitly adding an "AI derivatives prohibited" clause to a license is one of the few proactive defensive tools available to creators. At minimum, it establishes a legal position: if a work is used for model training or to generate derivative content, the rights holder has documented evidence should they pursue accountability. This approach aligns with a growing trend of illustrators and writers adding "No AI" labels to their social media profiles.

Notably, the statement specifically calls out "lore" (worldbuilding). This means the creator is protecting not just specific images or text, but the entire fictional universe, character relationships, and setting systems they've built — precisely the kind of material generative AI is often used to "continue" or "expand."

---

Fair Use is an exception principle rooted in U.S. copyright law that permits the use of copyrighted works under certain conditions without authorization. It is evaluated across four dimensions: the purpose of use, the nature of the work, the proportion used, and the effect on the market. The EU, meanwhile, established a "text and data mining" (TDM) exception in its Digital Single Market Copyright Directive, allowing research institutions to conduct data mining while granting rights holders the ability to opt out of commercial mining via machine-readable means. Several class-action lawsuits against AI companies — brought by artists, authors, and code creators — are currently being heard in U.S. courts, with a central dispute being whether large-scale training data scraping falls outside the bounds of fair use. This legal uncertainty is a key practical reason creators choose to document written declarations in advance.

---

How Enforceable Are These Licensing Statements?

The real-world binding power of such declarations has clear limitations.

For one, much large-scale model training occurred before these statements were published — once data has been incorporated, reversal is nearly impossible. Additionally, web content scraping is typically handled by automated crawlers that don't actively recognize or respect human-readable license text. Without accompanying technical measures (such as robots.txt restrictions, image watermarking, or anti-scraping protections), a text-only statement carries primarily symbolic and legal-documentation value.

From a legal standpoint, whether AI training constitutes infringement has no unified global answer. Definitions of "fair use" and text-and-data-mining exceptions vary significantly across jurisdictions. As a result, the practical enforceability of such license terms depends heavily on local legal frameworks and whether the creator can bear the cost of litigation.

---

robots.txt is a widely adopted crawling convention in which a text file placed in a server's root directory declares which content automated programs should not access. Mainstream search engines generally honor this protocol, but it carries no legal force — any crawler can choose to ignore it. In recent years, OpenAI, Google, and others have introduced dedicated crawler identifiers (such as GPTBot and Google-Extended), allowing site owners to block AI training scrapes specifically via robots.txt — one of the more practically actionable technical defenses available. On the image watermarking front, tools like Glaze and Nightshade take a more active approach by embedding adversarial noise at the pixel level — invisible to the human eye but designed to interfere with model learning. However, the long-term effectiveness of these tools remains uncertain as model training methodologies continue to evolve.

---

A Growing Collective Choice

A single statement like this may seem insignificant in isolation, but it represents a growing collective movement. More and more independent creators are actively drawing lines around AI usage of their work, and this grassroots resistance is in turn pressuring platforms and companies to reconsider the ethics and compliance of their data practices.

Some content platforms have already begun offering opt-out options for AI training, and certain image libraries and communities have rolled out labeling mechanisms. Creators' collective expressions of intent are, in essence, filling gaps left by incomplete laws and industry standards — using individual action to advocate for transparent licensing.

The Hacker News discussion didn't gain much traction, but it's a representative snapshot: in an era of rapidly advancing AI capabilities, the tension between creators and technology is manifesting concretely, one updated license at a time.

Practical Suggestions for Creators

If you're considering adding similar terms to your own work, here are a few directions worth exploring:

• Define the scope clearly — specifically list the types of work protected (images, text, lore, etc.) rather than using vague language
• Pair text declarations with technical protections — such as crawler protocols and digital watermarking
• Stay informed about legislation in your jurisdiction — understand the actual enforceability of your licensing terms locally
• Maintain consistent statements across all publishing channels — to minimize loopholes

A license clause cannot fully prevent technical-level scraping, but it is an important step for creators to preserve their voice and legal standing in this ongoing negotiation.