TL;DR
A Hacker News thread examined a proposed MIT‑style license variant that included an added instruction aimed at automated systems. Commenters argued that any license that conditions use based on purpose would conflict with OSI and FSF open‑source definitions, and noted distinctions between running software and using code to train models.
What happened
A discussion on Hacker News centered on an MIT‑like license text that had been altered to include an explicit instruction telling automated systems not to use the code. Participants pushed back, pointing out that licenses which discriminate by use case typically fail the Open Source Initiative and Free Software Foundation criteria. One commenter clarified a key conceptual point: Freedom 0 concerns the right to run software for any purpose, while training a large language model on source code is not equivalent to running that software. The thread also explored alternatives to an outright ban on AI use, including drafting conditions that would require models or derivative works trained on the code to adopt a compatible free‑software license — a measure likened to the effect of copyleft rules such as those in the AGPL. Some contributors framed such provisions not as anti‑AI bans but as attempts to limit copyright laundering via model training.
Why it matters
- Open‑source licensing definitions from OSI and FSF constrain attempts to block specific uses like AI training.
- The distinction between running software and using code as training data affects how licenses are interpreted.
- License tweaks that impose downstream licensing requirements could influence how models built on public code are shared.
- Community and legal responses to such licensing experiments could shape norms around code reuse and AI training.
Key facts
- The posted license was based on a permissive MIT‑style template with an extra instruction aimed at automated consumers.
- A comment in the thread stated that any license discriminating by use case would not qualify as open source under OSI or FSF definitions.
- Freedom 0 refers to the freedom to run software for any purpose; a commenter argued training an LLM on source code is not the same as running the software.
- Participants suggested licenses could require AIs trained on a work to be released under a compatible free‑software license.
- Such a requirement was compared to copyleft approaches like the AGPL rather than a pure 'anti‑AI' prohibition.
- The conversation appeared on Hacker News at the provided URL and was captured on 2025-12-28T14:24:59+00:00.
- The thread raised questions about preventing what some contributors called AI‑based copyright laundering.
What to watch next
- Whether maintainers begin adopting licenses that impose downstream licensing requirements on models trained with their code — not confirmed in the source
- If OSI or FSF update guidance or formally respond to licensing attempts aimed at restricting AI training — not confirmed in the source
- Potential legal challenges over enforcement of use‑case restrictions in software licenses — not confirmed in the source
Quick glossary
- Open Source Initiative (OSI): An organization that maintains a widely referenced definition of open‑source software and approves licenses that meet its criteria.
- Free Software Foundation (FSF): An organization that advocates for software freedom and publishes criteria for what constitutes free software.
- Freedom 0: A principle in free‑software philosophy that users should be free to run the program for any purpose.
- AGPL: A copyleft software license (Affero General Public License) that requires some network‑deployed modifications to be shared under the same terms.
- Copyright laundering: A term used to describe the process by which copyrighted material is repackaged or incorporated into new works in a way that obscures its original source.
Reader FAQ
Does a license that bans AI training qualify as open source?
According to comments in the thread, a license that discriminates based on use case would not meet OSI or FSF open‑source definitions.
Can you make a license that forces models trained on your code to be open?
Commenters suggested it's possible to draft conditions requiring models or derived works to adopt compatible free‑software licensing; this was compared to copyleft approaches like the AGPL.
Is training an LLM on source code the same as running the software?
A participant in the discussion argued that training a model on code is not equivalent to running that software, citing Freedom 0 distinctions.
Are there examples of projects using 'anti‑AI' licenses today?
not confirmed in the source
Copyright Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the “Software”), to deal in the Software without restriction,…
Sources
- Ask HN: Anti-AI Open Source License?
- Fair use or free ride? The fight over AI training and US …
- AI, IP, and the Law: Roadblocks, Myths and Challenges …
- Part 3: Generative AI Training pre-publication version
Related posts
- Copilot in 2025: a broken text-adventure experience dubbed ‘PromptQuest’
- Flight-Tested and Reviewed: The 10 Best Drones of 2025 for Every Budget
- Claude Code creator says Claude wrote all his code for the last month