Checking 241 Pan-African Review articles with Pangram: AI flags only spike after 2025, countering detector bias claims
menhguin · x · 2026-10-12
Responding to multiple articles claiming AI text detectors (especially Pangram) disproportionately flag African writers, Cremieux re-checked Pan-African Review's archive with Pangram 4.0:
- 150 articles from 2019–2023: 0% flagged as AI
- 2024: only 3.3% flagged
- 2025: 50% flagged; 2026: 89.7%, mostly as AI-generated or AI-assisted
The takeaway: detection isn't biased against African authors by race — it fires because AI use actually surged from 2025 onward. The author defends Pangram as the only working AI text detector that quickly adapts to humanizer tools, directly contradicting the mainstream discrimination narrative.
More from Models
- Dev observes models barely improved at one-shotting in 6 months — tool use and thinking are the ladder — rickasaurus · 2026-10-12
- Leaked Claude Test: One Prompt, Two Hours, a Full 3D Car Model — iamfakhrealam · 2026-10-12
- Redditor feeds 15 years of mental health crisis logs to Gemini, ChatGPT and Claude, results surprise the group — KrazyKreutz · 2026-10-12
- Prepping for Claude Pro: home broadband, $10 VPS, used router — but virtual cards fail — lxfater · 2026-10-12
- AWS open-sources Strands Decider 2B, a tiny low-latency decision model — gleech · 2026-10-12
- Decision model leaderboard called out for omitting GPT-6 Sol and Luna Decisions — gleech · 2026-10-12