Pangram Raises $9 Million to Detect AI-Generated Content

The Surge of Automated Content Demands an Urgent Technical Response

The internet is increasingly drowning in a rising tide of automatically generated content — SEO-optimized articles devoid of real value, fake posts, and disinformation campaigns powered by large language models. Amid this reality, Pangram, a company specializing in synthetic content detection, has emerged as one of the leading technical solutions. The company recently announced the completion of a $9 million funding round, alongside the launch of its latest models for detecting AI-generated text and images.

Funding Round Details and New Products

Menlo Ventures led the funding round, with participation from several investment funds including Haystack, ScOp, Script Capital, and Cadenza. The funding announcement coincided with the launch of Pangram 4, a new text detection model that exceeds 99% accuracy in identifying AI-assisted writing — whether fully generated or blended with human content. The company also unveiled an AI-generated image detection model currently in research preview, with a public release expected in the coming weeks.

How Does the Detection System Work?

The company was founded by Max Spero and Bradley Emi, both graduates of Stanford University's AI and machine learning program, approximately two years ago. Pangram's system works by training a large machine learning model on tens of millions of verified human documents, then generating a synthetic counterpart for each document that mimics the topic, length, and writing style — but is produced by modern language models.

Through this process, the model learns the subtle stylistic differences that distinguish machine-generated writing, without relying on metadata or hidden watermarks. Detection extends across multiple degrees of AI involvement, including text originally written by humans and subsequently edited by AI.

Why Is Demand for Detection Tools Growing?

Evidence points to a notable rise in AI misuse across various contexts:

• Lawyers who submitted AI-generated fabricated legal citations in court proceedings, resulting in fines and sanctions.
• Politicians who read AI-generated scripts verbatim in official speeches.
• The open-access research platform arXiv announced a strict policy banning researchers for a full year if clear signs of unreviewed language model outputs are detected prior to publication.

Spero argues that knowing whether content is human or machine-generated has become a necessity, not a luxury — as this knowledge fundamentally affects the level of trust a reader extends to the material in front of them.

How Can Pangram Be Used?

The company makes its tools available through multiple channels:

• A monthly subscription at $20 via the website.
• A Chrome browser extension that automatically detects the nature of posts on platforms such as X, LinkedIn, Substack, Reddit, and Medium, displaying the ratio of human to AI content.
• An API already integrated with prominent platforms including Substack and Quora, as well as universities, educational institutions, publishing agencies, and recruitment firms.

Real-World Performance Testing

In tests conducted by TechCrunch, the model successfully detected articles fully generated by tools such as ChatGPT and Claude, and manual paraphrasing attempts largely failed to fool it. The model also proved resilient against attempts to prompt language models into producing text designed to evade detection systems. However, the model was not without flaws — it misclassified some human-written sentences as AI-generated, which reflects the fundamental challenge facing the field of synthetic content detection to this day.

A Competitive Market in the Making

Pangram is not alone in this space; notable competitors include Winston AI, Originality.ai, Copyleaks, and GPTZero, each vying for a share of a rapidly expanding market. What sets Pangram apart is its technical approach based on deep stylistic learning, combined with extending detection capabilities to images alongside text — making it a comprehensive tool in the face of the growing synthetic content challenge.

✦ بقلم فريق دروب أيديا