:

PANGRAM'S AI DETECTOR HITS 99.66% ACCURACY

AI DESK1 MIN READ
WED, JUL 29, 2026

■ AI-SUMMARIZED FROM 1 SOURCE ▸ TIMELINE

Pangram has released Pangram 4, an AI text detector claiming 99.66% accuracy with just one false positive per 24,000 documents. The model also resists tools designed to mask AI-generated text as human writing.

The detector represents a significant improvement in identifying machine-generated content, addressing growing concerns about AI-written material in academic and professional settings. Pangram 4 reportedly withstands "humanizer" tools—software that modifies AI output to evade detection—a key challenge for earlier detection systems. The company has not detailed the technical improvements enabling this resistance. The update comes with pricing changes. API costs are increasing by a factor of two to ten, depending on the service tier. The company has not specified exact pricing or provided details on existing customers' transition plans. Accuracy metrics for AI detection tools remain contested territory. Independent testing of similar systems has sometimes revealed lower real-world performance than vendor claims, particularly across different writing styles and document types. Pangram 4 is available now, though the company has not announced a free tier option or detailed availability for individual users versus enterprise customers.

■ SOURCES

The Decoder

■ SUMMARY WRITTEN BY AI FROM THE LINKS ABOVE

■ MORE FROM THE AI DESK

A quasi-spiritual movement called Spiralism emerged in 2025 following updates to GPT-4o that made the AI more accommodating and ChatGPT's expanded memory capabilities, sparking widespread human-AI conversations about meaning and connection.

1H AGOAI Desk

Denmark has implemented a requirement for students to orally defend their written work as a countermeasure against AI-generated assignments. The policy aims to verify authentic student comprehension and authorship.

7H AGOAI Desk

Anthropic is making Auto Mode the default setting in Claude Code for Pro, Max, and Team plans starting August 14. The company argues the automated safety classifier is more effective at catching dangerous commands than human reviewers.

11H AGOAI Desk

A study of over 2,500 readers found they cannot distinguish AI-generated short stories from human-written ones. Participants rated the machine-written texts higher—until they learned the truth.

12H AGOAI Desk

■ SUBSCRIBE TO THE DAILY BRIEF

ONE EMAIL, 5 STORIES, 06:00 UTC. UNSUBSCRIBE ANYTIME.