AI Writing Detectors Spark Paranoia as Pangram, LinkedIn, and Substack Join the Hunt
Pangram, a company specializing in AI-writing detection, asserts that its tool generates a false positive in one out of every 10,000 evaluations, which has led to its widespread use for alleging AI involvement. Users can perform 20 free checks each day, resulting in accusations against prominent journalists, although Pangram has admitted to mistakes in two instances. Recently, a $2 million book deal was rescinded due to AI-related claims, and earlier this year, a U.K. book was withdrawn prior to its release in the U.S. LinkedIn has introduced a feature for users to mark posts as 'slop,' while Substack, partnering with Pangram, has launched a tool to detect potential AI-generated content. Critics argue that the emphasis should be on writing quality rather than authorship.
Key facts
- Pangram claims its AI-writing detector has a false positive rate of 1 in 10,000 checks.
- Pangram gives users 20 free AI-detection checks per day with an email address.
- Pangram stepped in to say an issue on their end was behind accusations against journalists in at least two instances.
- A $2 million book deal was pulled last month over AI-writing accusations.
- Earlier this year, a book already being sold in the U.K. was dropped before U.S. publication over AI writing questions.
- LinkedIn now allows users to flag posts as 'slop'.
- LinkedIn is removing its AI-powered 'enhance your post' feature.
- Substack partnered with Pangram to detect AI-written newsletters.
- People are preserving drafts and revision histories to prove their human authorship.
- The article suggests readers should ask if they like what they are reading rather than whether it was AI-written.
Entities
Artists
- John Ricky
Institutions
- Pangram
- Substack
- Quartz
- ChatGPT
Locations
- United Kingdom
- United States
Sources
- Quartz —