Pangram raises $9 million for tools that spot AI-made text and images

In short: Pangram, a New York startup, raised $9 million and released a new version of its tool that tries to tell whether content was written by AI.

What happened

Pangram said it raised $9 million to grow its AI detection software. The round was led by Menlo Ventures, with Haystack, ScOp, Script Capital, and Cadenza also participating.

At the same time, Pangram launched a new text detection model called Pangram 4. The company says Pangram 4 is over 99% accurate at finding AI-assisted writing, including writing that mixes human and AI work, and it can better spot “humanizer” tools (software that tries to rewrite AI text so it looks more human).

Pangram also introduced an AI image detection model called Pangram Image. It is only available in a research preview for now, and the company says it plans a wider release in the coming weeks.

Pangram sells access through a $20 per month web subscription and a Chrome extension that labels posts as you browse sites like X, LinkedIn, Substack, Reddit, and Medium. It also offers an API, which is a way for other companies to plug Pangram into their own products, like adding a “check this” button.

Why it matters

More text and images online are being made with AI, which can make it harder to know what to trust. Tools like Pangram aim to act like a “nutrition label” for content (a quick hint about what you are looking at), which could help readers, schools, publishers, and platforms decide when extra caution or disclosure is needed.

Source: TechCrunch AI