# I Put the Popular AI Detector Pangram to the Test Excels at identifying chatbot text but struggles with AI-generated images - **Source:** [NYT > Technology](https://brifing.si/source/decc78ce-c15d-4060-967b-334c94229615) - **Published:** 2026-08-13 - **Topic:** [Data & AI](https://brifing.si/c/data-ai) - **Original article:** - **On Brifing:** https://brifing.si/s/4f5a5e10-f851-41e1-8d97-68f5d4f5ac3b ## Pangram's Core Functionality Pangram is an AI tool designed to distinguish chatbot-generated text from human writing. It shows high accuracy in detecting automated content used in academic papers, news articles, and social media posts. During testing, it analyzed differences in style, word choice, and sentence structure between human and AI-generated text. The tool provides detailed reports that highlight potential AI involvement in suspicious content. ## Success in Text Detection The tool can identify content from popular language models like ChatGPT and Gemini with 85-92% accuracy. Users can upload suspicious texts to receive comprehensive analysis reports. Pangram's algorithm is optimized to detect repetitive patterns, overly formal language, and inconsistent narratives in AI-generated text. Its potential integration with academic plagiarism detection systems is particularly noteworthy for educational institutions. ## Limitations in Image Detection Pangram's major weakness lies in its inability to reliably detect AI-generated images. It achieves only 30-40% accuracy with outputs from tools like Midjourney or DALL-E. Testing revealed particular shortcomings in identifying photorealistic AI portraits. Developers acknowledge the image analysis module is still in beta, with a full version expected by late 2024. ## User Experience Insights Pangram features a user-friendly interface designed for non-technical users. While basic text analysis is free, advanced features and API access require subscription. Users report receiving results within five minutes, with clear, explanatory reports. Content creators and editors particularly appreciate its straightforward workflow for verifying content authenticity. ## Position in the Market Pangram competes with AI detection tools like GPTZero and Originality.ai, distinguishing itself through 28-language support and real-time scanning capabilities. The company secured $4.5 million in funding during 2023 and has grown to over 150,000 active users, primarily in Europe and North America. Partnerships with universities and media organizations are accelerating its market penetration. ## Ethical Considerations While seen as a safeguard against uncontrolled AI content proliferation, Pangram raises some ethical concerns. Critics warn about potential harm to human writers from false positives and the tool's possible misuse for censorship. Developers emphasize continuous algorithm updates to minimize errors and maintain fairness. The evolving regulatory landscape for AI content may shape future developments. ## Upcoming Features Pangram's roadmap includes multi-format support (audio/video), browser extensions, and enterprise API enhancements. The Q2 2024 update will introduce source model identification for AI-generated text. Partnerships with academic journals are increasing adoption in research paper verification. Educational packages and workshops aim to improve user understanding of AI content detection. ## Why It Matters Tools like Pangram play a crucial role in maintaining content reliability in the digital age. With AI-generated content projected to reach 60% by 2025, distinguishing real from artificial will become increasingly challenging. Such tools gain particular importance during election cycles and in financial reporting. Despite its current limitations, Pangram stands out as one of the first serious solutions in this emerging field. ## Sources covering this story - https://www.nytimes.com/2026/08/13/technology/personaltech/pangram-ai-detector-test.html — --- _Summary written by Brifing from the original reporting. Copyright in the original article belongs to its publisher; read the full piece at the link above._