Publishers lose attribution the moment their text is ingested by an AI. Encypher solves that - cryptographic provenance, embedded invisibly in the text itself, that survives copy-paste, wire services, and aggregators.

Co-Chair, C2PA Text Provenance Task Force  ·  Patent Pending  ·  Open Source Core

When AI systems train on or reproduce publisher content, the original author and outlet disappear. There is no technical mechanism to prove origin after the fact - no metadata survives copy-paste, no EXIF equivalent exists for text. The result is willful infringement at scale, with publishers unable to detect it, quantify it, or enforce against it.

EncypherAI embeds a cryptographically signed, invisible payload directly into Unicode characters. The signature travels with the text, no matter where it ends up.

graph TD
    A["c2pa-org/c2pa-spec<br/>Open Standard"] -->|implements| B["encypherai/c2pa-text<br/>Reference Implementation<br/>Python · Go · Rust · TypeScript · AGPL-3.0"]
    B -->|Python core| C["encypherai/encypher-ai<br/>Python Library · PyPI<br/>Unicode variation selector signing"]
    C -->|powers| D["Private Commercial Suite<br/>Enterprise API · Dashboard<br/>WordPress · Chrome · Office SDKs"]

graph TD
    OSS["Open Source - AGPL-3.0<br/>Anyone can implement the standard"]
    API["Free Hosted API<br/>Zero friction adoption"]
    COM["Commercial - Private<br/>Attribution analytics<br/>Formal notice generation<br/>Evidence packages · Batch ops<br/>SSO · Revocation"]

    OSS -->|"everyone implements the standard"| API
    API -->|"enforcement requires"| COM

from encypher.core.unicode_metadata import UnicodeMetadata

# Embed verifiable provenance into any AI-generated text
signed_text = UnicodeMetadata.embed_metadata(text, private_key, payload)

# Verify - survives copy-paste, wire services, and aggregators
result = UnicodeMetadata.verify_text(signed_text, public_key)
# -> {'verified': True, 'payload': {...}, 'timestamp': '...'}

uv add encypher-ai
# or
pip install encypher-ai

Regulatory frameworks around AI transparency are advancing globally, with major jurisdictions now requiring disclosure of AI-generated or AI-manipulated content. The C2PA standard - the technical backbone EncypherAI implements - has been adopted by leading AI platforms, browser vendors, and camera manufacturers. Publishers and media organizations are actively looking for a standards-compliant way to assert and track provenance.

Developers - Drop-in provenance for any text pipeline. pip install encypher-ai · Open an issue

Publishers & Platforms - If you're evaluating text provenance at the enterprise level, reach out on LinkedIn.

Researchers & Standards Contributors - The reference implementation lives at encypherai/c2pa-text. C2PA spec work happens at c2pa-org/c2pa-spec.

Solo builder - I've written every line of the Encypher stack, from cryptographic core to commercial dashboard. Eagle Scout. FIRST Robotics alumnus. I think a lot about open-source sustainability, standards governance, and what it means to build infrastructure that the internet actually trusts.