You know how many folks have been up in arms regarding potential copyrighting issues with ChatGPT?
Enter….
This!
“Tom Goldstein
@tomgoldsteincs
Jan 25
#OpenAI is planning to stop #ChatGPT users from making social media bots and cheating on homework by “watermarking” outputs. How well could this really work? Here’s just 23 words from a 1.3B parameter watermarked LLM (large language model). We detected it with 99.999999999994% confidence. Here’s how 🧵…”
The entire thread is at
twitter.com/tomgoldsteincs/status/1618287665006403585
but the thing to consider?
Truly, never use AI output without first spinning it to sound like *you*.
The chances of it being detected will become *dramatically* less.
Enjoy!
23 words out of a 1,300,000,000 watermarked