Pangram has raised $9 million in a funding round led by Menlo Ventures, with participation from Haystack, ScOp, Script Capital, and Cadenza, TechCrunch reported on July 29, 2026. At the same time, the company launched Pangram 4, its latest tool for spotting AI-written text, and Pangram Image, a tool for spotting AI-made images. The image tool is currently available as a research preview, with a wider release planned in the coming weeks.

Pangram says Pangram 4 is over 99% accurate at identifying AI-assisted writing and content that mixes human and AI work. It is also built to catch "AI humanizers" — tools that rewrite AI-generated text so it looks like a person wrote it. The company was founded about two years ago by Stanford AI and machine learning graduates Max Spero and Bradley Emi, after the launch of ChatGPT.

The detection system works by training a large machine learning model on tens of millions of documents known to be written by humans. For each of those documents, an AI text generator creates a matching synthetic version. The model then learns the differences in style between the human writing and the AI version — the kinds of word choices and sentence patterns that tend to appear in AI output but not in human writing. The tool does not rely on hidden tags or watermarks. It reads the writing style itself.

Instead of giving a simple "human or AI" answer, Pangram reports how much AI assistance was involved. Co-founder Max Spero has said that using AI is acceptable as long as the writer discloses it, which positions the product as a transparency tool rather than something designed only to catch people out.

The company's blog tracks a regular pattern of updates that reflects an ongoing back-and-forth in this field. Pangram published results on July 23, 2026 saying it detects Anthropic's Claude Opus 5 with 99.82% accuracy across 1,107 tested examples. On July 9, it confirmed it could detect OpenAI's GPT 5.6 on the same day that model was released. Earlier posts documented detection of Claude Fable 5 on June 9 and GPT-5.4 on March 6. A June 24 post called "AI Has an Eye for AI" compared 30 AI detection tools for accuracy, speed, and false positive rates, noting that OpenRouter credits were supplied by Pangram to fund the research.

Pangram's own data gives a picture of how common AI content has become online. A July 9 blog post reported that a scan of over one million social media posts found AI-generated content on every platform checked, with one in three top LinkedIn posts flagged as AI-generated. A May 15 post said that 67% of people consuming online content say they are spotting misleading information from AI. A May 21 post noted that a Commonwealth Short Story Prize finalist was accused of producing AI-generated work.

The need for these tools reaches into education and academic publishing. Pangram published a guide on May 12 explaining how schools can add tools like its detector to Google Classroom through plugins. Separately, the open-access research archive arXiv introduced a policy in 2026 that can trigger a one-year submission ban for authors who fail to review AI-generated content in their papers, addressing problems such as made-up references and leftover instructions from prompts.

Looking at the bigger picture, launching text and image detection at the same time points to a shift. As AI gets better at producing convincing images and text, detection tools that only cover one format are incomplete. Pangram Image entering research preview, even before it is widely available, is a sign that detecting AI text alone is no longer enough.

There is a deeper tension in this field that is worth understanding. Every time Pangram announces it can detect a new AI model, that announcement is also evidence that the model was already being used to produce content before the detector caught up. The cycle is built into the structure of the problem: an AI model is released, the detector is updated to catch it, tools emerge to disguise the AI output, and then the detector is updated again. Pangram 4's claim of defeating AI humanizers is the latest round in that cycle, not an end to it.

Pangram has also taken part in policy conversations. A 2023 blog post titled "Statement on Biden's AI Safety Executive Order" argued that the government should fund AI detection research, not just focus on regulating the AI models themselves. That argument came before the current funding round but fits with the company's direction: building detection as a lasting piece of infrastructure, not just a one-off feature.

The $9 million round, while small by AI-era funding standards, is enough to grow a company whose main assets are a trained model and a research team that tracks new AI releases almost in real time. Five firms participated, with Menlo Ventures leading, which suggests investors believe the demand for detection tools is durable rather than a short-term reaction to today's AI models.

Whether that demand holds depends on how the online content world develops. If people settle on clear norms for disclosing AI use, or if watermarking standards become widely adopted, the market for detection tools shrinks. If neither happens, and AI-generated content keeps flooding social media at current rates, tools like Pangram's become a standard part of the infrastructure for any platform, publisher, or school that needs to know whether a given piece of content was written by a person.