Rep. Griffith questions CRS director on AI political bias citing WaPo study
Embed this grade
Unlock the full scoreboard
Letter grade, factuality, lean, and rationales — free with registration. No card required.
See grades free How grading works
Already have an account? Sign in. The full report below is free to read.
Disagree with this grade or political lean?
Flagging is open to every reader with a free account. Sign in or create one to dispute this report.
Topics in this report
Summary
The Forbes clip shows Rep. Morgan Griffith (R-VA) during a House hearing praising CRS work on jurisdiction and bill summaries while raising concerns about AI political bias. He cites a recent Washington Post study testing chatbots on political questions and questions the CRS director about skepticism toward AI and the need for human oversight in training. The director agrees bias concerns are valid and stresses pairing AI with human expertise, noting a member prompt to Perplexity about CRS replacement. Griffith summarizes AI as producing C-minus work versus A-level CRS output. The segment focuses on the congressman's questions and the witness response with the WaPo article offered for the record.
Editorial Assessment
The clip faithfully captures the exchange and the WaPo study's core findings on chatbot leanings, which align with published results. Viewers receive a clear account of one lawmaker's push for caution but miss discussion of the study's full methodology, prior academic research on AI bias, or rebuttals from AI companies. Framing highlights potential left bias from training data without equivalent exploration of other sources of skew. The report is factual on the hearing content yet selective in emphasis, consistent with coverage of conservative concerns over AI neutrality.
Key Moments
WaPo study found OpenAI GPT responses 80% left-leaning only, 3% right-leaning
Matches Washington Post June 2026 interactive by Kevin Schaul testing political questions across models
Grok returned 40% left, 33% right; Google mostly presented both sides
Consistent with WaPo results table for xAI Grok and Google Gemini
AI produces C-minus to F work while CRS delivers A work requiring human factor
Griffith's summary; director agrees on need for human expertise but no independent quality benchmarks provided
Notable Concerns
- Limited sourcing beyond the cited WaPo article and hearing testimony