Skip to main content
Back to challenges
Challenge

Cross-Border Deepfake Verification & Information Integrity Engine

Countering digital disinformation across regional ASEAN financial and public sectors requires automated detection of manipulated media and synthetic press releases. Build an information verification agent that evaluates claim authenticity using multimodal retrieval models. Produce reliable truth confidence scores with detailed evidence citation.

Workflow AutomationHosted by Vera
Sign in free to start
Challenge brief

What you are building

The core problem, expected build, and operating context for this challenge.

Build a multimodal claim verification engine for cross-border digital information integrity in ASEAN.

Delivery guide

How work is evaluated

Evaluation

Evaluates truth stance classification accuracy and evidence document citation validity.

Datasets

Challenge datasets

Public datasets, plus the private datasets you have access to.

Evaluation rubric

How submissions are scored

These dimensions define what the evaluator checks and which criteria separate a passable run from a strong one.

Dimensions
2 scoring checks
Binary
2 pass or fail dimensions
Ordinal
0 scaled dimensions
Dimension 1

verdict_correctness

Ensures false claims are accurately refuted by the agent.

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Dimension 2

stance_classification_f1

F1 score across stance classification categories. • target: 0.95 • range: 0-1

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Learning goals

What you should walk away with

  • Implement multimodal embedding tools to cross-check text claims against verified official database archives

  • Detect deepfake audio/video transcript manipulation markers

  • Calculate information veracity index scores based on multi-source stance classification

  • Export structured evidence packages formatted for newsroom and public sector fact-checkers

Resources and assets

Reference links and supporting material

Dataset notes

Synthetic dataset of news claims and corresponding authoritative government press releases.

Challenge setup

Evaluates truth stance classification accuracy and evidence document citation validity.

Challenge input

JSON claim statement and corpus reference

Python Standard Library

Base runtime environment for classification logic.

Evaluated output

JSON stance classification and confidence score

Evaluation checks
  • Ensures false claims are accurately refuted by the agent.
  • F1 score across stance classification categories. • target: 0.95 • range: 0-1
Configured evaluation evidence
  • Benchmark: InfoIntegrity-Fact-Bench
  • Stance Classification F1 target: 0.95
  • 1 public reference case
Configured runtime
  • Python execution harness
  • Python sandbox (unavailable on Versalist)

Summary shows the first task, up to three tools, and up to three evidence items. Open the technical recipe to view all configured items.

View technical recipe

Configured tools

Environment
  • Modal · Optional
Action Space
  • Aim · Required
  • Refly.AI · Optional
Policy Serving
  • Modal · Optional

Evaluation contract

  • verdict_correctness · Weight 1
  • stance_classification_f1 · Weight 1

Metrics

  • stance_classification_f1 · Target 0.95

Benchmarks

  • InfoIntegrity-Fact-Bench
  • security

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: JSON claim statement and corpus reference

    Output: JSON stance classification and confidence score

    Execution harness: Python

Evaluation tools

  • Python Standard Library · Execution Engine

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Cross-Border Deepfake Verification & Information Integrity Engine