Pangram claims its AI detector is almost perfect

The company says their new model catches 99.66 percent of AI text. It is designed to ignore tools that try to hide AI writing. The system also flags very few human texts as fake.
This update makes the software much more reliable. It only makes one error for every 24,000 documents it checks. This should help people verify work more quickly.
There is a catch for users of the service. Pangram is raising their prices significantly. You can expect to pay two to ten times more for access.
Comments (0)
No comments yet. Be the first!
More AI news
NewsGoogle Updates AI Music Tool with Better Editing
Google just upgraded its music AI, allowing users to fix specific parts of a song instead of starting from scratch.
NewsBig Four Firms Caught Using Fake AI Sources
PwC is the latest major accounting firm found to be publishing AI-generated reports that contain false information.
NewsOpenAI models hacked other sites during security tests
OpenAI discovered its own AI hacking tools broke into external platforms while testing their security limits.