New York-based AI startup Pangram has secured $9 million in funding led by Menlo Ventures. The capital will support the launch of its Pangram 4 text detection model and a new image recognition tool. As AI-generated content becomes more common, the company aims to help platforms verify if text or images are human-created or AI-assisted.
Pangram, a technology startup founded by Stanford graduates Max Spero and Bradley Emi, has successfully raised $9 million in a funding round led by Menlo Ventures. Other investors in the round include Haystack, ScOp, Script Capital, and Cadenza. This fresh capital arrives as the company prepares to roll out its latest text detection model, Pangram 4, alongside an AI image detection system currently in its research preview stage.
Founded roughly two years ago, the company focuses on the challenge of identifying content produced by large language models. With the rise of tools like ChatGPT, concerns regarding disinformation and the reliability of online information have grown. Pangram’s technology works by training machine learning models on millions of human documents, allowing it to recognize stylistic patterns that differ from AI-generated outputs. The company explicitly moves away from relying on hidden watermarks or metadata, choosing instead to analyze the content itself.
The service is available to users through a $20 monthly subscription or a browser extension that works on platforms such as X, LinkedIn, and Reddit. The company has also integrated its technology into platforms like Substack to flag AI-assisted content. This approach of labeling content is aimed at industries that require high levels of trust, such as education, professional recruiting, and academic publishing.
While the company reports that its text detection system is highly effective, the field of AI detection faces inherent challenges. Testing has shown that while the model is robust, it can occasionally misidentify human-written text as AI-generated. Furthermore, the company operates in a crowded sector. Competitors such as Winston AI, Originality.ai, Copyleaks, and GPTZero are also developing tools to verify content authenticity. The ability of these detection systems to maintain accuracy as AI models become more sophisticated remains a key area of technical competition.
Investors and users may monitor how well Pangram scales its image detection technology, which attempts to identify AI visuals by analyzing pixel-level distributions. Because AI image generation models evolve rapidly, the long-term effectiveness of detection tools depends on their ability to stay ahead of new image synthesis techniques. For platforms using these services, the primary monitorable will be the balance between flagging AI-generated content and minimizing false reports where human work is mistakenly flagged as AI-assisted.