LoRes AI Policy

v1.0.0, Oct 2026

Goals

The primary goals of this policy do not relate to the code quality of the LoRes project, or our efficiency in delivering features (although we do care about those things). Instead, our primary purpose here is to join the resistance. To us this means:

• Hastening the inevitable popping of the AI bubble as quickly as possible, by refusing to support the business models of big AI
• Rejecting the social license of rapid data-centre expansion by not contributing to their digital workloads or revenue.
• Reducing AI normalisation in software development by fostering a community that encourages and supports coding by humans, at all skill levels

We are specifically focussing on fighting the fight that aligns most with this project, by pushing back on the climate destruction and centralisation of power caused by big AI. We appreciate other projects with different policies fighting back against other AI harms.

Non Goals

We are in a moment of rapid change across many industries, happening as the climate crisis is impacting lives across the globe, and fascist political movements are attempting to find people to blame and attack. The fight for a better world is against tech oligarchs and nationalist warmongers, not other workers and volunteers building projects with different views and strategies to us. As such, it is not our goal to:

• Call out developers, designers and other workers for their views on “AI” in general, or for their use of it in other projects. This policy applies to this project only.
• Create a space where discussion must conform to ideological purity.
• Use pejoratives to describe other workers as a strategy to remove AI social license (sloppers, clankers, etc).

Scope

This policy applies to contributions to any LoRes managed project, which is any software listed at https://lores.tech/contributing/software/.

It applies to all kinds of contributions, including source code, pull requests, images, documentation, bug reports and so on. It does not apply to your own research or other processes you use prior to making a contribution.

Policy

❌ Banned

This project does not accept contributions which contain any part that was:

1. Generated by a LLM with even one of the following characteristics:
   1. Hosted, operated or sold by a tech company that aims to eventually return profit, or provide share value to, investors or share-holders. This absolutely includes, but is not limited to: OpenAI, Anthropic, Google, Meta, SpaceXAI, Amazon, Microsoft.
   2. Run in a data centre, whether by an AI provider, cloud provider, or “self-hosted”
   3. Run on anonymously rented consumer hardware, such a mesh of LLMs
2. Not fully understood by the submitter.
3. Any use of LLMs of any sort in communication intended for humans (excluding code). This means no AI in user documentation, website copy, scripts for talks or videos, etc.

✅ Allowed

Providing the above is met, this project does accept contributions which contain any of:

• Use of locally run spelling or grammar checkers, even if they use closed-weight LLMs, providing they run locally.
• The use of any assistive technology you need to to make contribution to this project accessible to you, including screen readers, translation tools, and assistive interfaces. This is allowed even if these tools use or contain LLMs. While we recommend open-weight, locally run models for these tools, we do not require them.
• Use of auto-complete models to assist with in-editor predictive typing, providing they are open-weight and either:
  1. Run locally on your development computer, or on consumer hardware already in use in your home/office, or:
  2. Meet the conditions for agentic/chat models below.
• Use of LLM models in an agentic/chat context to generate code, providing all of the following conditions are met:
  1. Run on a computer physically in your local region (within 20kms of where you usually work on the project)
  2. Run on renewable electricity
  3. Is an open-weight model
  4. Is operated either by the individual developer, an educational institution, a non-profit, or a worker co-operative (or similar structure that shares profits, agency and decision making amongst all it’s workers/members).

What about harms this policy doesn’t address?

To read more about our reasoning, see our blog post introducing this policy which covers some of the rationale, and discusses some of the areas we have not addressed.

This AI policy is not designed to be suitable for all organisations, it specifically targets areas relevant to this project.