@CryptPad what is your LLM policy?
Enquiring minds would like to know.
0 Boosts
0 Likes
This federated comment is not yet available here.
@kauer
Not really.. CryptPad's funding is just paying people's salaries, no more. It still needs subsidies to exist. https://blog.cryptpad.org/2026/02/18/CryptPad-Funding-Status-2026/ My time and some of other XWiki employees is not even counted in the CryptPad budget. If you consider that paying employees is "benefits" then yes sure.. You can judge us, but do it based on proper information.
@CryptPad @abucci @onepict
@ldubost @CryptPad @abucci @onepict You are so missing the point. I am saying that there are no benefits of any description to you or the project that can justify using generative AI, a hugely wasteful and destructive technology based entirely upon theft.
I get that you would prefer not to hear that.
In 2025 I can see that project members were not interested in AI
https://forum.cryptpad.org/d/2447-ai-in-cryptpad
But also with a kind of we don't have time to review.
At a time when many of us are looking for alternative tools that are explicitly not developed with LLMs, it would be nice to have a formal policy.
Hello @onepict
Thanks for your interest in CryptPad and its development.
So far we don’t have any AI-policy, however as a NLNet-funded project, we are following the NLNet AI policy 1.1: https://nlnet.nl/foundation/policies/generativeAI/1.1/
It allows for responsible and transparent use of AI-tools without degrading quality, clarity and maintainability. It has been recently updated to a 2.0 version only allowing read-only use of AI-tools. CryptPad is still compliant with it.
For the sake of transparency, CryptPad is using AI for security analysis ahead of releases to catch oversights, but the fixes are written and reviewed by our developers.
We are planning to make a survey to our end-user and developers to know their stance about the use of AI-tools in the development and plan to publish soon a temporary AI policy to clarify how it has been so far and our future plans.
On another note, the forum posts you linked to refers to AI-features in the product. Due to the way AI is working nowadays, we have no plan so far to integrate AI features in CryptPad in the foreseable future as it will go against the principle of end-to-end encryption.
@CryptPad thank you for taking the time to answer this question.
@onepict you’re welcome!
As we said in a previous message, we described our position in a blogpost: https://blog.cryptpad.org/2026/11/10/Temporary-AI-Policy/
@CryptPad thank you for being so thorough and open about this in your blog post.
I did wonder how much your fiscal home would affect Cryptpad development so this clears it up.
@onepict@chaos.social
From their blog post:
From their blog post:
While as citizens we are concerned about the risks associated with AI usage and negative externalities, both at our level and for society in general, we also have to evaluate the positive benefits AI can bring to small teams and to our end goals.This is alarming. It suggests they are not clear about their values in the project, which is one of the vulnerabilities being exploited by AI boosters.
We believe we need to keep an open mind and study all aspects before deciding on a final policy.
@abucci With the current choices by Libreoffice and Colabora to use LLMs, it is concerning.
Especially to those of us who are extremely concerned by the ethical implications and also just don't want LLM produced code in our lives.
It is being forced in everywhere.
@onepict@chaos.social Oof, I wasn't aware that Colabora was embracing LLMs too (I don't use it or follow the project).
Somewhat tangentially it seems to me we're seeing an uptick in projects stating "We have no AI features in the software but use AI to produce code, tests, bug reports, ...". That seems to be where a bunch of folks who have concerns about AI but don't want to stop using it are settling.
Somewhat tangentially it seems to me we're seeing an uptick in projects stating "We have no AI features in the software but use AI to produce code, tests, bug reports, ...". That seems to be where a bunch of folks who have concerns about AI but don't want to stop using it are settling.
@abucci @onepict Yeah it's so disingenuous. "No malware misfeatures in the user facing software" should not even be something to brag about just a lowest possible bar. The actual concern is codeslop in the code we're expected to run, introducing vulnerabilities, risking the long term free status of the software, driving off all the principled contributors and turning the dev process into a nazi bar.
