Generative AI production based solely on text prompts—even the most detailed—is not protected under current copyright law, the U.S. Copyright Office says. 🤖📜

The department issued this guidance in an extensive report on policy issues concerning AI, focusing on the copyright protectability of various AI-generated products. The paper concludes that although generative AI is a new technology, existing copyright principles can be applied without changes to the law—and these principles offer limited protection for many types of works. 📚✨

New guidelines indicate that AI prompts currently do not provide sufficient control to "make users of a system "AI systems themselves cannot own copyrights.) This holds true whether the prompt is extremely simple or involves long strings of text and multiple iterations. No matter how many times a prompt is revised and resubmitted, the final output is always the same. reflects the user's acceptance of the system's interpretation of AI, rather than being an author of the expression it contains," the report says. 📝🎨

This decision appears to rule out protections for works such as "Théâtre D'opéra Spatial," a Midjourney image generated and awarded with controversy whose creator fought for a long time to register it with the Copyright Office. 🎭🖼️

The office illustrates the unpredictability of AI systems with an image Gemini produced of a cat smoking a pipe and reading a newspaper, noting that Gemini ignored some prompt instructions and added a few things in its own way — including an “incongruous human hand.”

Comparing this process to Jackson Pollock's splatter painting method, where he did not control the exact placement of paint on the canvas, but “controlled the choice of colors, the number of layers, the depth of texture, and the placement of each addition to the overall composition—and used his own body movements to execute each of these decisions.” Ultimately, the office writes: “the problem It is the degree of human control, rather than the predictability of the outcome.” 🎨🙌

At the same time, the Copyright Office says that simply using AI to assist in human creative production does not jeopardize that work’s ability to receive legal protection. There’s a difference between using AI as a tool to assist in a creative work and “AI as a replacement for human creativity,” and the office says further analysis is required. But it assures creatives that using AI to outline a book or generate song ideas should not affect the copyrightability of the final human-produced work, since the author simply “references, but does not incorporate, the output.” 🎶📖

Artists can gain some protection if they feed their own work to an AI system for modification — for example, by using a tool to add 3D effects to an illustration. AI-generated elements of the work would still not be protectable, but if the product original remains recognizable, the “perceptible human expression” in the work could be covered by copyright. 🖌️💫

Individuals may also receive protection for works that incorporate AI-generated content as long as there is a significant creative modification. A comic book with AI-generated images may be covered if a human organizes those images and pairs them with human-created text, although individual AI-generated images would not be protected.

Similarly, “a film that includes AI-generated special effects or background art is copyrightable even if the AI-generated effects and art separately are not.” On a “case-by-case determination,” even images generated from prompts could be copyrightable if a human selects and remixes specific areas of the image. The office compares these scenarios to the creation of derivative works of human-created art, but without the human creator. 🎥🖼️

A separate question is whether text prompts themselves can be protected by copyright. In general, the office compared prompts to “instructions” that convey unprotectable ideas, but acknowledged that some particularly creative ones might include “expressive elements.” This, however, does not translate into the work they produce being protected. 🤔📝

The Copyright Office did not rule out the possibility that this could change if the technology evolves. “In theory, AI systems could one day allow users to exert so much control over how their expression is reflected in output that the system’s contribution would become routine or mechanical,” the report says. But for now, it doesn’t appear that prompts “adequately determine the expressive elements produced, or control how the system translates them into output.” 🔮💻

This document is part of a larger effort by the Copyright Office to clarify policy questions and identify legal loopholes around AI, beginning with a July 2024 report encouraging new laws on deepfakes. The office plans to soon issue a third and final report on its findings regarding “the legal implications of training models.” of AI with works protected by copyright.” 📆🔍