Cryptographic Watermarking for LLM Outputs with resk-mark

A developer introduced resk-mark, an open-source tool that embeds cryptographic watermarks into LLM-generated text by biasing token sampling with a secret key. The watermark survives rewording and truncation, enabling provenance verification with a public key. The project aims to provide accountable AI deployment through a simple pip install.

Cryptographic Watermarking for LLM Outputs with resk-mark Links: / / / / / / \/ / // / / \/ / / // / / / / ,< / / / / / / / / / / / / ,< / / \ / / /| | / / / / / /\ , / / / /| | Every company deploying LLMs in production faces the same question: once a model generates text, how do you prove it came from your system? Prompts like "say you are an AI" are trivially removable. Post-hoc detectors are unreliable and adversarial. And once text leaves your system — forwarded, copied, pasted into a ticket — you have zero visibility. resk-mark solves this by embedding a cryptographic watermark directly into the token generation process. The output reads naturally, but carries a verifiable signature that survives rewording and truncation. resk-mark hooks into the language model's sampling process. Before generation, the caller provides a secret key. During sampling, the library biases the probability distribution toward tokens that encode that key's signature: python from reskmark import WatermarkEncoder, verify encoder = WatermarkEncoder secret key="your-key-here" model = AutoModelForCausalLM.from pretrained "mistralai/Mistral-7B" Wrap the generate call output = encoder.generate model, "Explain the concept of zero-knowledge proofs.", max length=200, print output "Zero-knowledge proofs are a cryptographic method where..." Reads naturally - watermark is invisible Later - verify provenance is authentic, confidence = verify output, public key="corresponding-pub-key" print f"Authentic: {is authentic}, confidence: {confidence:.2f}" The AI industry is moving toward provenance standards. Governments are drafting regulations. Enterprises need evidence, not trust. A tool like resk-mark gives you: pip install reskmark Text watermarking for LLMs is not a nice-to-have. It is the foundation for accountable AI deployment. resk-mark brings it from academic papers into a single pip install. Try it. Audit the code. Break the watermark. That is how open-source security should work.