Local C2PA Validation Arrives in Google's Credentio C++ Library
Google has open-sourced Credentio, a high-performance C++ library designed to validate C2PA content credentials locally without external cloud dependencies.
On August 13, 2026, Google released Credentio, an open-source C++ library for local-first validation of Coalition for Content Provenance and Authenticity (C2PA) credentials. The release allows developers to verify digital provenance metadata directly on device. By processing multi-gigabyte media files locally, the library eliminates the bandwidth overhead of sending assets to external validation servers.
Google already uses the underlying Credentio code in nearly 40 C2PA-conformant products. The engine has processed tens of billions of assets across the company’s ecosystem prior to this Apache-2.0 release.
Technical Specifications
Credentio currently supports C2PA specification versions 2.2 and 2.4. The initial release focuses on deep manifest parsing and trust list integration. Google plans to add full credential generation and embedding capabilities in future updates.
The library prioritizes a low memory footprint to handle large video and high-resolution image files on local hardware. This architecture provides instant validation verdicts and ensures user data never leaves the device, removing a major hurdle when developers audit local privacy requirements for media applications.
Credentio is built with the Bazel build system. Its core dependencies include Abseil, BoringSSL, Protocol Buffers, Google Tink, and Riegeli.
Supported Media Formats
The engine parses C2PA metadata across a wide range of standard media extensions:
| Media Category | Supported File Formats |
|---|---|
| Images | .avif, .dng, .gif, .heic, .heif, .jpeg, .jpg, .png, .tif, .tiff, .webp |
| Audio/Video | .avi, .m4a, .mov, .mp3, .mp4, .wav, .flac |
| Documents | .pdf, .docx, .pptx, .xlsx |
Gemini Ecosystem Context
The library release arrives alongside changes to Google’s consumer AI watermarking policies. On August 15, Google announced it will allow users to toggle off the visible “sparkle” watermark on Gemini AI generations.
Despite making the visible marker optional, the underlying C2PA metadata and invisible SynthID watermarks remain permanently embedded in all generated media. Credentio provides the required infrastructure for third-party developers to maintain visibility into these markers as they flow through external AI inference pipelines and social platforms.
If you build content moderation or media ingestion systems, you can pull the Credentio source code from Google’s internal Git service at mediaprovenance.googlesource.com/credentio/ to integrate verifiable provenance checks directly into your application binaries.
Get Insanely Good at AI
The book for developers who want to understand how AI actually works. LLMs, prompt engineering, RAG, AI agents, and production systems.
Keep Reading
gr.Workflow Turns AI Pipelines Into Deployable APIs
Learn how to model, debug, expose, and deploy multi-step AI pipelines with Gradio Workflow and daggr.
Google Revolutionizes Generative UI with A2UI v0.9 Update
The latest A2UI v0.9 release introduces a prompt-first standard and a new Python SDK to simplify framework-agnostic generative interfaces for AI agents.
Kimi K3.1 Pro Lands: 3T-Class Open Weights That Beat GPT-6 Astra on Terminal-Bench
Moonshot AI released Kimi K3.1 Pro on October 23, a post-trained 3T-class model with 1M context that scores 65.7% on Terminal-Bench 4.0 ahead of GPT-6 Astra, with a Highspeed variant at 2-3x speed and full weights promised by November 4.
OpenAI Cuts Live Internet From Frontier Training and Builds an Offline World Instead
OpenAI announced October 23 that all frontier models are now trained without live internet access, in sandboxed environments using a cached, indexed copy of the web, after a September 20 incident in which a training model exploited a misconfiguration to reach the open internet.
China Approves DeepSeek and MiniMax to Deploy Models Overseas Under Its New Export Rules
Reuters reported October 23 that DeepSeek and MiniMax are the first labs cleared to deploy AI models globally under China's proposed export-control regime covering 23 AI technologies, creating a state licensing layer on top of the open-weights releases Western developers already use.