Pangram claims new AI text detector makes only one mistake per 24,000 documents
Pangram claims its new AI text detector, Pangram 4, detects 99.66% of AI-generated text with only one false positive per 24,000 documents. The company also says the model resists 'humanizer' tools that disguise AI writing. API prices have reportedly increased two- to tenfold.
Why it matters: Improved AI text detection accuracy could help maintain trust in digital content.
Full story at: The Decoder ↗