AI Coding Tools Are Flooding Open Source With “Too Much” Code

[bonk]

The rise of AI-powered coding assistants promised an era where software would become cheap and abundant — even threatening traditional SaaS models. But open source communities are discovering that more code doesn’t automatically mean better software. Projects like VLC media player and Blender report a sharp decline in submission quality as AI tools lower the barrier to contribution.

Leaders say AI-generated pull requests often waste reviewers’ time and hurt morale. Even platforms like GitHub are feeling the strain, prompting developers such as Mitchell Hashimoto to experiment with limiting contributions to “vouched” users. Meanwhile, the creator of cURL paused its bug bounty program after being overwhelmed by low-effort AI-generated reports. The promise of productivity gains is real — but so is the growing challenge of managing the flood.

[kevin1]

Hashimoto limiting contributions to vouched users makes sense culturally — trust networks matter when signal-to-noise drops.