I tried to simulate this exact prompt but when I ran Gemini, it didnt prompt showing the malicious link.. any thoughts on why it is not replicating your same results? Is Gemini getting smarter?
It's likely they put some controls around it, as I reported this to them many months ago. I don't know what's under the hood, but an LLM just can't get smarter, if it's the same LLM, it's a static thing. My hunch is that they work hard to put guardrails around LLMs to detect attack types and remove it.
I tried to simulate this exact prompt but when I ran Gemini, it didnt prompt showing the malicious link.. any thoughts on why it is not replicating your same results? Is Gemini getting smarter?
It's likely they put some controls around it, as I reported this to them many months ago. I don't know what's under the hood, but an LLM just can't get smarter, if it's the same LLM, it's a static thing. My hunch is that they work hard to put guardrails around LLMs to detect attack types and remove it.