Expose AI Citation Fakes Instantly
LLMs bluff. You get blamed. Laoshu.ai slams the brakes on fake citations so you ship only the truth.
One slip, and your credibility's gone.
- Catch fake citations before they catch you.
- Ship with confidence. No more AI lies.
- Open-source (AGPL), zero black box.
- Self-hosted, your keys, your data.
- Paste any output from Perplexity/ChatGPT Deep Research.
- Laoshu fetches every source.
- Get instant pass/fail verdicts.
⚠️ Warning: This is an early proof-of-concept release. It more or less works with Perplexity outputs.
Run Laoshu.ai with Docker:
docker run -p 3000:3000 -e SCRAPINGANT_API_KEY=your_key -e OPENAI_API_KEY=your_key mikulskibartosz/laoshu:0.2.0-devOpen your browser at http://localhost:3000, drop text, get the verdict.
Bring your own ScrapingAnt & OpenAI keys.
Contributions are welcome! For now, we're accepting feature requests and bug reports, but not code contributions yet.
- Feature requests: Open an issue and describe what you'd like to see
- Bug reports: Include steps to reproduce and expected vs actual behavior
- Documentation: Typos, clarifications, and improvements are appreciated
We'll open up code contributions once the core functionality is more stable.
And once there are "we", for now it's a single person project ;)
Source code: https://github.com/mikulskibartosz/laoshu
Built by Bartosz Mikulski — AI hallucination-prevention specialist. Need custom fixes? Let's talk.
Lǎoshǔ.ai is released under the GNU Affero General Public License v3.0. Commercial entities wishing to use it under a different license should contact me.