# I Let Pangram Loose on My Gmail Inbox. Here’s How Much AI It Found

- URL: https://talkfuse.com/story/i-let-pangram-loose-on-my-gmail-inbox-heres-how-much-ai-it-found/
- Published: 2026-11-09T10:00:00+00:00
- Updated: 2026-11-09T10:36:24+00:00
- Original publisher: Wired
- Original article: https://www.wired.com/story/i-let-pangram-loose-on-my-gmail-inbox-heres-how-much-ai-it-found/
- Topics: AI, Technology

## Key points

- Pangram now scans Gmail for signs of AI-written messages.
- It classified 8 of 108 tested threads as AI.
- The small test cannot establish overall accuracy.

## Summary

Pangram has added a Gmail integration that scans incoming messages for signs of AI-written text, letting users label, archive or filter emails based on its assessment. The feature matters as AI-written messages become harder to distinguish from human writing, though users must trust Pangram with access to email contents.

The integration requires Pangram’s $20-a-month individual plan and scans the 100 most recent threads, then new messages; the company says email contents are not retained or used for training. In a test of 108 threads, it marked eight as AI, 37 as human and the rest as mixed. The author’s known human-written emails and a Copilot-generated message were classified correctly, but the sample was small, short emails may go unclassified, and only the first message in each thread is checked.

---
Summarised by TalkFuse (https://talkfuse.com/). Quote the original publisher for the reporting and link this page for the summary and discussion.