I built AI Flag Bot to tackle the growing problem of AI-generated accounts flooding X (Twitter).
How it works: - Spot a suspicious post? Reply with @AIFlagBot flag to report it - The community votes on whether it's AI-generated - Votes are weighted by voter reputation, which goes up or down based on accuracy - Once consensus is reached, the account gets added to a shared blocklist - Paid subscribers get automatic blocking synced to their X account ($5/m to cover API costs)
The reputation system is the core idea — voters who consistently identify AI accounts correctly gain more influence, while bad actors lose it. Think of it as a community immune system.
Stack: Rails 8.1, SQLite (Solid Stack), Solid Queue for background jobs, Kamal for deployment. The X API integration handles mention polling, OAuth, and block syncing with careful rate limit management.
It's early but the model is working. Would love feedback on the approach — especially whether community consensus can scale as a detection method.