CREdible
Decentralized Media Verification Platform powered by CRE and AI
What it is
Credible
Decentralized Media Verification Infrastructure
The Problem
We live in an era where AI can generate photorealistic images and videos in seconds. Deepfakes are indistinguishable from reality. Screenshots get passed off as original footage. Manipulated media spreads faster than corrections. Journalists, activists, legal teams, and everyday people have no reliable way to prove that a photo or video is authentic — that it was captured by a real camera, on a real device, at a real moment in time.
At the same time, content creators have no privacy-preserving way to receive support for the work they produce. Publishing your wallet address publicly exposes your entire financial history on-chain.
What Credible Is
Credible is a decentralized media verification platform powered by Chainlink CRE (Compute Runtime Engine) that cryptographically proves the authenticity of digital content. The platform consists of:
- Browser Extension — Verify images and videos from anywhere on the web (YouTube, Twitter, IPFS, direct URLs)
- Mobile Camera App — Capture and verify photos/videos at the moment of creation
- Verification Oracle — Chainlink CRE workflow that analyzes content, generates cryptographic proofs, and records results on-chain
- Public Registry — Immutable on-chain record of all verified content, queryable by anyone
- Private Donations via Chainlink ACE — Every piece of verified content carries a shielded address to protect donor and creator’s identity and adhere to compliance rules.
How It Works
Browser Extension
- Content Submission Users paste a URL to any image or video they want to verify. The extension supports:
- YouTube videos
- Twitter/X posts and media CDN URLs
- IPFS-hosted content (Pinata, ipfs.io, etc.)
- Direct image/video URLs (.jpg, .png, .mp4, etc.)
The extension captures both the content URL and the source page URL (for attribution).
- Chainlink CRE Processing The URL is submitted to a Chainlink CRE workflow that:
- Fetches the content from the provided URL
- Hashes the content using SHA-256 to create a unique cryptographic fingerprint
- Analyzes with Google Gemini AI to detect signs of:
- AI generation or synthesis
- Digital manipulation beyond standard photography
- Screen recordings vs. authentic captures
- Deepfake characteristics
- Uploads to IPFS via Pinata in a structured format:
- record/ content.jpg (or .mp4) — original media file metadata.json — hash, analysis, timestamp, source
- Writes to blockchain — Content hash + IPFS CIDs recorded in an Ethereum transaction on Sepolia testnet
- Verification Result The extension displays:
- Full AI analysis of authenticity
- Content hash (SHA-256)
- Blockchain transaction hash (Etherscan link)
- IPFS storage URLs for content and metadata
- Source attribution (which website the content was found on)
All results are stored in a JSON database that feeds a public website where anyone can browse verified content.
Camera App
- Capture & Hash When a user takes a photo or records a video in the Credible camera app, the media file is immediately hashed using SHA-256 directly on the device before it ever leaves the phone.
- IPFS Upload The app uploads the original media and metadata to IPFS via Pinata, generating an immutable content-addressed record.
- CRE Verification The IPFS CID and hash are submitted to the same Chainlink CRE workflow, which:
- Fetches the content from IPFS
- Analyzes it with Gemini
- Verifies the hash matches
- Records everything on-chain
This creates proof that the content existed at a specific point in time and was captured by the Credible app, not edited later.
The Verification Chain
Every verification produces an unbroken chain of cryptographic proof:
Content source (web URL or device capture)
↓
SHA-256 hash computed
↓
IPFS — content + metadata stored immutably
↓
Chainlink CRE — Gemini AI analyzes authenticity
↓
Analysis uploaded to IPFS
↓
Blockchain transaction — hash + IPFS CIDs recorded on Sepolia
↓
Anyone can verify: hash the content → query the registry → retrieve proof
Why Chainlink CRE?
Traditional verification systems rely on centralized servers that can be compromised, taken offline, or manipulated. Credible uses Chainlink CRE as decentralized verification infrastructure:
- Trustless Timestamping — Multiple Chainlink DON nodes reach consensus on when content was verified
- Censorship Resistant — No single entity can take down the verification record
- Cryptographic Attestation — Each verification is signed by the Chainlink network
- Immutable Record — Blockchain ensures the verification history can never be altered
- Decentralized AI Integration — CRE acts as a bridge between Web2 AI services (Gemini) and Web3 verification infrastructure
What Makes Credible Different
Not just metadata stamping — Many systems only record that "a file existed." Credible actually analyzes the content using AI and stores the full analysis on-chain.
Universal verification layer — The same CRE workflow verifies both:
- Content captured directly in the camera app (strong proof of origin)
- Content found on the web (verification of current state + AI analysis)
Public, queryable registry — Anyone can look up a content hash and retrieve its full verification record from IPFS without needing to trust Credible.
Decentralized by design — Verification happens on Chainlink DON nodes, storage happens on IPFS, records live on Ethereum. No centralized points of failure.
ACE Compliant Private Transactions — Allows donors and content creators to receive donations anonymously to protect their identity.
Use Cases
For journalists and investigators: Prove that footage was captured authentically and hasn't been manipulated. Verify content from sources before publishing.
For content platforms: Integrate Credible verification as a trust layer. Display verification badges on authentic content.
For legal and compliance teams: Establish chain of custody for digital evidence with cryptographic proof.
For everyday users: Verify viral images and videos before sharing. Combat misinformation at the source.
For creators: Prove authenticity of original work. Establish provenance for digital art and photography.
The Tech Stack
- Chainlink CRE — Decentralized compute layer for running verification workflows
- Hashing — CryptoJS SHA-256 on-device and in CRE
- Google Gemini AI — Multi-modal analysis of images and videos
- Private Payments Chainlink ACE — shielded addresses + off-chain private transfers
- IPFS (Pinata) — Content-addressed immutable storage
- Ethereum Sepolia — Blockchain for permanent verification records
- Browser Extension — User interface for web content verification
- Mobile App — Cross-platform mobile camera app
- Express.js Backend — Aggregation layer for website display
The Vision
Credible isn't just a verification tool — it's open infrastructure for media provenance. The CRE workflow acts as a universal verification oracle that any application can call:
- News organizations can integrate Credible verification into their CMS
- Security cameras and dashcams can submit footage for timestamped proof
- Social media platforms can display verification badges
- Legal evidence systems can establish chain of custody
The verification record lives on IPFS and Ethereum forever, independent of Credible's existence. That's the power of decentralized infrastructure.
How it Works
How It's Built
Credible is built across five layers that work together end-to-end.
Browser Extension The extension is built with vanilla JavaScript and runs in Chrome. When a user submits a URL, the extension captures both the content URL and the source page URL for attribution, then POSTs it to the Chainlink CRE workflow via HTTP trigger. It polls the Express backend until a verification result is stored, then displays the full analysis — AI verdict, SHA-256 hash, IPFS links, and Etherscan transaction link — directly in the extension popup.
Mobile Camera App (React Native / Expo) Built with Expo and react-native-vision-camera. When a user captures media, the app reads the raw file and computes a SHA-256 hash using CryptoJS before the file ever leaves the device. It collects device metadata, camera settings, and EXIF data, then performs a two-step upload to IPFS via Pinata — first uploading the content to get a content CID, then uploading a structured folder containing the media and a metadata JSON file with the hash embedded. Every payload also carries a shielded donation address for private payments.
Chainlink CRE Workflow The core verification oracle. Triggered via HTTP, the workflow fetches or receives the content URI, passes it directly to Google Gemini for AI authenticity analysis — detecting AI generation, pixel manipulation, screen recordings, and deepfake characteristics — then uploads the full analysis JSON to IPFS via Pinata's pinJSONToIPFS endpoint. Finally it writes the content hash and analysis CID to the CredibleRegistry smart contract on Ethereum Sepolia using EVMClient and the runtime.report() signing pattern. Results are also posted back to the Express server for website display.
Smart Contract (Solidity) The CredibleRegistry contract is deployed on Ethereum Sepolia. It stores an immutable mapping of bytes32 content hashes to IPFS analysis CIDs and emits a Verified event on each write. Anyone can query the contract with a SHA-256 hash to retrieve the on-chain verification record — no trust in Credible required.
Private Donations (Chainlink ACE) Built on Chainlink's Automated Compliance Engine. The creator registers their wallet and receives a shielded address — a privacy-preserving Ethereum address that cannot be linked to their real identity without the server's secret key. This shielded address is embedded in the content's IPFS metadata and stored on the Express server. On the website, donors connect MetaMask, sign an EIP-712 typed data request, and POST directly to the Chainlink ACE API. The transfer is validated off-chain against an on-chain PolicyEngine and credited to the creator's private balance — with no public on-chain footprint linking donor to creator.
Links
Created by
- Roberto Iturralde