New York-based Pangram raises $9M to combat AI-generated content, launches Pangram 4 (99% accurate text detection) and Pangram Image (research preview), amid growing demand for distinguishing human vs. AI content.
Substack integrated Pangram's AI detection software, allowing users to scan content for AI-generated text. This move aims to build trust and combat "AI slop," though it might initially expose AI use in newsletters.
Substack partnered with Pangram to launch an AI detection tool for content over 100 words, available now on web/iOS, with Android support coming. It also added a creator disclosure space. The announcement included a dig at LinkedIn and acknowledged detection limitations.
Der KI-Detektor Pangram verspricht 99,98% Treffsicherheit bei der Erkennung maschinell generierter Texte. Doch was misst es wirklich und wie zuverlässig sind seine Urteile in der Praxis?
KI-Detektoren wie Pangram versprechen hohe Treffsicherheit bei der Unterscheidung von menschlichen und maschinell erstellten Texten. Doch die Programme erkennen Stilmerkmale, nicht die tatsächliche Urheberschaft. Dies führt zu Fehlinterpretationen und der Gefahr, menschliche Texte fälschlich als KI-generiert einzustufen.
Analysis suggests parts of Pope Leo XIV's encyclical on AI may have been written by AI, with 40-100% of certain paragraphs flagged as AI-generated
AI Detector Developer Pangram Labs’ browser extension tagged several posts from the Pope’s X account.
