tldraw pauses external contributions, citing rise in low-quality AI code and pull requests

tldraw pauses external contributions, citing rise in low-quality AI code and pull requests

tldraw, the open source collaborative whiteboard project, is suspending all external pull requests due to increasing concerns about the quality of AI-generated submissions. According to an updated “Contributions policy” on their GitHub repository, the team has announced they will automatically close all pull requests from external contributors. This temporary measure will remain until GitHub provides better technical controls over who can contribute, when, and where.

While tldraw will stop accepting external code contributions, users can still file issues, report bugs, and participate in discussions. The project's maintainers note a sharp rise in pull requests written entirely or largely by artificial intelligence tools. Although some submissions are technically correct, many fail to demonstrate adequate understanding of the codebase or lack meaningful engagement with the maintainers after submission.

Following clarification from the tldraw team, they stated that the presence of AI in itself is not the main problem. Instead, they question whether AI-assisted external code contributions are currently valuable, given observed quality and workflow issues. The decision has been positively received by contributors and users alike, who appreciate the team's concern for the stability and quality of the project.

by Paul

nblackburn
ba
ddnnsoul1472
nblackburn found this interesting
tldraw icontldraw
  21
  • ...

tldraw is a collaborative digital whiteboard accessible at tldraw.com. It features an intuitive editor and user interface, with libraries available on npm, allowing seamless integration into your product. Key features include a drawing board and digital drawing capabilities, all without requiring registration. Rated 4.5, tldraw is a versatile tool for creating a drop-in whiteboard experience.

Comments

Augusto Goulart
0

Yep, can confirm, Lua's mailing list has just seen a wave of LLM generated "vulnerability" reports. Maybe this is related? It takes less than five seconds to see how those so-called vulnerabilities make no sense (also, using markdown in a mailing list is a crime on its own). If your project is getting dozens of these everyday, then I would see why the maintainers would want to limit contributors.

SleipnirTheHorse
0

Gee, AI is actually fairly new untested technology which is incredibly expensive and does not have that many real world applications yet, who knew? s/

barmunk
3

Is there anything AI bros won't ruin?

RDF0909
1

Wish my company would adopt that policy. Half our new devs can't explain their code. We have a couple foreign teams from India and since AI the quality has gone from bad to horrible. But the shareholders are happy so that's all that matters (so I'm told).

Gu