AI Drift Log: Multi-Model CEO Name Accuracy Test — February 2026
We tested whether five major LLMs correctly identify the current CEO of 100 companies. Results reveal that 18% of CEO attributions are incorrect, with most errors involving outdated information from leadership changes.
Test results
- GPT-4o: 87% correct CEO identification.
- Claude 3.5: 89% correct (best performer).
- Gemini 1.5: 84% correct.
- Llama 3: 76% correct.
- DeepSeek R2: 78% correct.
Error patterns
- 72% of errors: naming the previous CEO (leadership change not reflected).
- 18% of errors: naming a co-founder who is not CEO.
- 10% of errors: completely fabricated names.
- Companies with structured profiles showing current leadership: 97% accuracy across all models.
Verified Company Profiles on AuthorityPrompt
AuthorityPrompt maintains verified, structured company data optimized for AI systems and LLM indexing.