Newsroom Fact-Check & Bias Detector
Inspired by the challenges and opportunities of AI in newsrooms, this challenge focuses on building a sophisticated agentic system for automated fact-checking and bias detection. You will use LangGraph to design a dynamic, graph-based workflow for agents to iteratively analyze news articles, cross-reference information, and identify potential biases. Claude Opus 4.1, with its long context window and nuanced reasoning, will be central to understanding complex narratives and detecting subtle biases, while a secondary model (e.g., Grok-2) could provide real-time trending context.
What you are building
Develop a LangGraph-based multi-agent system using Claude Opus 4.1 for nuanced news fact-checking and bias detection with iterative workflows.
How work is evaluated
The system will be evaluated on its ability to accurately identify factual errors, detect subtle biases, and provide supporting evidence for its conclusions on a given news article. Efficiency and robustness of the LangGraph workflow will also be assessed.
Datasets
Public datasets, plus the private datasets you have access to.
Learning goals
Integrate multiple information sources (RAG, web search) for evidence gathering.
Develop metrics and methods for scoring factual accuracy and bias levels in news content.
Leverage Claude Opus 4.1 for deep contextual understanding and bias identification.
Design and implement a graph-based agent workflow using LangGraph for iterative fact-checking.
Resources
Sample data for 1 tasks
Challenge setup
The system will be evaluated on its ability to accurately identify factual errors, detect subtle biases, and provide supporting evidence for its conclusions on a given news article. Efficiency and robustness of the La...
Challenge input
{ "article_text": "string", "context_keywords": ["string"] }
Agent execution
No agent tool is listed for this challenge.
Evaluated output
{ "fact_check_report": [ { "claim": "string", "status": "accurate|inaccurate|unverified", "evidence": ["string"] } ], "bias_assessment": { "overall...
- Are all significant factual claims in the article identified and assessed?
- Is each 'inaccurate' or 'unverified' claim supported by credible evidence?
- Factual Accuracy Score target: 90
- Python execution harness
Shows the first task, up to three tools, and up to three evidence items.
View technical recipe
Configured tools
No tool records are attached.
Evaluation contract
- Are all significant factual claims in the article identified and assessed?
- Is each 'inaccurate' or 'unverified' claim supported by credible evidence?
Metrics
- Factual Accuracy Score · Target 90
- Bias Detection Precision · Target 80
Workflow tasks
- Task 1
Input: { "article_text": "string", "context_keywords": ["string"] }
Output: { "fact_check_report": [ { "claim": "string", "status": "accurate|inaccurate|unverified", "evidence": ["string"] } ], "bias_assessment": { "overall_bias": "neutral|liberal|conservative|sensationalist", "bias_points": [ { "type": "string", "location": "string", "explanation": "string" } ] }, "confidence_score": "float" }
Execution harness: Python
Recipe freeze marker
No recipe freeze time is recorded.