Worried about AI taking over your inbox? A new tool called Pangram can now scan your Gmail to flag emails potentially written by artificial intelligence. This comes as AI-generated content floods everything from menus to movies, making it harder to distinguish human work from machine output.
Pangram, known for its AI text and image detection, has launched a browser extension that integrates directly into Gmail. It analyzes incoming messages to determine if they're AI-generated. However, this feature isn't free; you'll need at least Pangram's $20-per-month individual plan to use it.
Setting up Pangram in Gmail is straightforward. After connecting your account via authorization screens, the tool begins scanning your 100 most recent email threads and all new messages going forward. Pangram assures users that email content isn't stored or used for training its models. While a degree of trust is involved, the plugin can be removed at any time.
Once scanned, Pangram's dashboard shows a breakdown: how many emails are suspected AI, likely human, or a mix of both. These categories can be applied as optional labels within Gmail. For those wanting to keep AI content at bay, the tool offers options to automatically archive or mark AI-written emails as spam. You can also exclude specific senders, domains, or labels from scans.
In a test of 108 scanned threads, Pangram flagged eight as AI-generated, with press releases being the most common culprits. Thirty-seven emails were labeled as human, and the rest as mixed. Notably, none of the emails from friends, family, or colleagues known to be human-written were misidentified. Tests sending emails composed by Microsoft Copilot (correctly flagged as AI) and the author themselves (correctly flagged as human) also proved the tool's accuracy, within its operational limits. Pangram does have a minimum email length requirement for detection and only scans the first email in a thread.
The author found the background scanning convenient and a potentially useful long-term tool, noting that knowing a message's origin can influence how it's perceived. The integration also opens the door to replying with AI-generated content if desired.
