AI detection startup Pangram has expanded its platform to identify AI-generated images and videos, introducing new tools aimed at helping users distinguish authentic visuals from synthetic content. The New York-based company, known for its AI-generated text detection service, now allows paid subscribers to upload videos and assess whether artificial intelligence was used to create or edit them. Its video detector can distinguish between partially AI-modified clips and entirely AI-generated videos, while providing a heatmap highlighting potentially altered sections. Pangram has also upgraded its image detection technology. Its Pangram Image 1.0 model is available through the company’s Chrome extension, which can scan images while users browse websites. According to Pangram, its image detector identified more than 99 per cent of AI-generated images during testing, with a reported false-positive rate of 0.27 per cent. The company also reported more than 98 per cent accuracy for its video detection tool in internal testing. These figures are based on the company’s own evaluations and should not be treated as independently verified performance guarantees. The technology faces challenges, particularly when analysing low-resolution images and photographs processed by smartphone cameras. Pangram says images below 512 × 512 pixels receive a reduced-confidence warning. Its CEO, Max Spero, has also highlighted the difficulty of distinguishing AI-generated visuals from genuine photographs enhanced through AI-based camera processing. The new image and video detection capabilities are available to paid subscribers. Users on the free tier can test a limited number of images without purchasing a subscription. As AI-generated content becomes increasingly common online, detection tools are emerging as one approach to identifying synthetic media, although their results still require careful interpretation. 

AdvertisementThe Puranic