←back to thread

302 points simonw | 2 comments | | HN request time: 0.591s | source
Show context
TacticalCoder ◴[] No.41891994[source]
That is amazing! I tried something not dissimilar, but only with a single picture: I took an AI model (I think it was GPT 4o, don't remember as I'm using several) and asked it:

    "Analyze this screenshot and tells me if everything seems legit"
And it could the difference between a legit screenshot and a phishing attempt. Without me telling it which site name was legit or not.

And it was pretty detailed: "There's an 'l' in interactlvebrokers.ie" that is made to look like an 'i'". Or something like that.

We're already getting and we're going to get lots* of shiny new helper tools.

replies(1): >>41893410 #
1. maeil ◴[] No.41893410[source]
One of the huge challenges is preventing false positives. Did you also try giving it a legit screenshot?
replies(1): >>41893603 #
2. bambax ◴[] No.41893603[source]
False positives are one thing, but the real danger are false negatives when someone trusts the model blindly. I'd rather examine a website with my own eyes than ask an LLM what it thinks.