All the Patents
Defensive Prior Art Through Exhaustive Claim Combination
Generating every possible combination of patent claims to establish comprehensive prior art -- preventing future patents based on recombination of existing ideas.
How It Works
Collect Patents
Gather all patents that have ever been filed, creating a comprehensive dataset of existing intellectual property.
Extract Claims
Parse individual claims from each patent, isolating the specific inventive elements that define each patent's scope.
Analyze in Vector Space
Using LLM-backed vector embeddings, place all claims in vector space and cluster them by semantic similarity.
Generate Combinations
Produce every possible combination of existing patent claims, establishing exhaustive prior art for the public domain.
Defensive Prior Art
If anyone in the future tries to recombine an existing claim into a new thing, they can point to our dataset as prior art: "They already did that. You can't patent that again, because it already exists as prior art."
This project was inspired by the successful All the Music project, which computationally composed over 400 billion melodies and released them into the public domain via Creative Commons Zero (CC0).
Why This Matters
AI and Patent Overload
AI systems can generate countless patent variations, threatening to overwhelm the patent system with questionable claims based on trivial recombinations.
Raising the Bar
By pre-emptively establishing prior art through exhaustive combination, the project raises the bar for what qualifies as a truly non-obvious invention.
Protecting Innovators
Genuine inventors benefit when the patent system filters out trivial recombinations, ensuring patents reward truly novel contributions.