Bans protect overwhelmed maintainers
Slop PRs are cheap to produce but expensive to review, and AI has destroyed the old signals of effort, so a ban works as a proof-of-work filter that prevents burnout.
Banning AI is lazy or self-defeating
The real issue is low quality, not AI, and outright bans drive away contributors, invite lying, and leave projects without AI-powered bug and vulnerability hunting.
Outside PRs are losing value
If maintainers can prompt an LLM themselves, external AI-generated PRs add little; issues, good prompts, or even token money may become the real contributions.