‘It’s terrifying’: WhatsApp AI helper mistakenly shares user’s number
‘It’s terrifying’: WhatsApp AI helper mistakenly shares user’s number
www.theguardian.com
‘It’s terrifying’: WhatsApp AI helper mistakenly shares user’s number
12 comments
-
-
I hate Facebook with a furious seething passion and have no vested interest in defending any of their shit, but this truly sounds like it made up a random number, with it being a total coincidence that the actual number belonged to another whatsapp user.
-
An LLM cannot generate random numbers. It has to pull from a list of numbers its model was built to include.
-
I mean the latter statement is not true at all. I'm not sure why you think this. A basic GPT model reads a sequence of tokens and predicts the next one. Any sequence of tokens is possible, and each digit 0-9 is likely its own token, as is the case in the GPT2 tokenizer.
An LLM can't generate random numbers in the sense of a proper PRNG simulating draws from a uniform distribution, the output will probably have some kind of statistical bias. But it doesn't have to produce sequences contained in the training data.
-
a WhatsApp user and whose number appears on his company website.
Either the number was in the training set, or it came up on the Web search somehow.
-
-
Shhhhh if someone doesn't know at thisbpoint, they can't be told.
-
-
12 comments
That's where the LLM found the number. It being a valid WhatsApp number is a coincidence.