A Note to Trust and Safety Teams from the Founders
Click to openBefore Pangram, I worked at Google and Yelp where I saw the dangers of fraudulent content first hand. Being AI researchers, we believe in AI being a force for good in the world and aim to prevent the misuse of AI in distorting reality and truth in places where content needs to be trusted.
We started Pangram here first and believe that AI-generated content flooding the web is one of the biggest issues of our time. When we started, we sought out to build the most accurate detector on the market, and through our research breakthroughs have trained the highest accuracy classifier with the lowest false positive rate.
We are a force multiplier for Trust and Safety teams to scale their initiatives at a low cost. Pangram powers teams with the tools they need to confidently set policies and effectively moderate AI content -- to preserve human generated content and flag AI generated content for company policy. We've created Pangram with you in mind and hope that it is a useful tool in combating the danger of AI content in reviews and user-generated content. We look forward to hearing any feedback.
Sincerely,
Max
Co-Founder of Pangram Labs
AI-written reviews erode trust and hurt the reputation of platforms.
The Federal Trade Commission has announced rules banning fake reviews, including AI-generated reviews.
AI-written reviews are not only unhelpful to consumers but also fool ranking systems, which can rank AI reviews highly due to their long-winded text and correct grammar.
U.S. Copyright Office
March 2023
100x
cheaper than doing it by hand.
Pangram helps scale Trust and Safety Teams for a fraction of the cost.
Human content moderation is an expensive and time consuming task.
Pangram is the only AI detector that outperforms human experts at identifying AI content.
Pangram helps automate this workflow, supporting time constrained safety teams whilst reducing costs.
Pangram's model can even detect AI in short form content.
Short social media posts, DMs and comments are often too short for traditional AI detectors to catch.
Pangram's model detects AI content as short as 75 words.
We can scan multiple pieces of content at the same time to provide a better signal on such short-form content.
Pangram's Technology Edge
Pangram's technology was developed by AI researchers from Tesla and Google with a commitment to exceptional accuracy. Our proprietary detection system actively learns from its shortcomings and improves itself when detecting AI content. It is this novel research method that has driven us to outperform other detectors in the market on reliability -- resulting in a 99.99% accuracy and the lowest false positive rate in the market.
Read our technical report here.
to our updates
