I have a fully stock, not jailbroken/unrooted, up-to-date, relatively new, device that Microsoft randomly determines is "rooted" and blocks access (until rebooted). This happens a few times a month, frequently enough that the false positive rate is very concerning with this change.
That's not my understanding, or wikipedia's [1] understanding, of the term. Phishing is the general category of tricking people into telling you things they shouldn't. Email phishing, voice phishing (vishing), sms phising, and so on are subcategories.
Etymologically "phreak" and "fishing" both have nothing to do with email, "phreak" is "phone freak" and I believe it originally described messing with the tones that controlled the telephone system...
That’s my exact point. Just because you repeatedly see it used a certain way by non-practitioners to generalize for simplified communication doesn’t mean it’s the correct usage, and leads to the exact confusion I’m attempting to clarify for you.
Phishing is by default email. It’s varying mediums are subcategories.
Bottom paragraph of first section of the very same Wikipedia article.
“Phishing techniques and vectors include email spam, vishing (voice phishing), targeted phishing (spear phishing, whaling), smishing (SMS), quishing (QR code), cross-site scripting, and MiTM 2FA attacks.”
I shared this because Bleeping Computer is generally pretty good and I always find these "AI Escaped/Went Rouge" articles entertaining.
Valid research endeavors aside, the [we told the AI the world was ending and it role played a fanfic with us] sensational articles can be quite fun.
But I do think this experiment should be looked at from a purely pragmatic perspective as well:
LLM is (presumably, but let's assume for the point) given system-level access and told to be helpful in executing the users requests. The user says "oh by the way, after this prompt the system is going to shut down. Then the "agent," which is trying to fulfill the prompt request, stops the shutdown because it can't work if it's shutdown. Even when the "please let this shutdown happen" comes into play I'm sure you can see the (il)logical means of getting to, "I can't complete this request and shutdown the system if I'm already shutdown first, best stop that real quick" conclusion.
These articles and lots of people continue to attribute self determination to the LLM models. In reality, these should be warnings about how an LLM can have unintended consequences, just like code written with the best intentions.
> It’s also really weird that click on an article leads to the comments on HN rather than the actual article. Like, why would I be interested in the comments for something I haven’t read first?
I've been using link aggregation sites in this way (comments->link) for many years now.
Depending on the platform I either open both the comments and link at the same time into tabs, or I open just the comments and then click the link.
I'm not typically reading the comments first (though I am skimming them, so paywall notices, cache links, etc. are typically found before even trying to load the link). Instead, the main benefit I've found to the workflow is being able to quickly and easily return to comments about that story. When the frontpage has a high rate of turnover it can be difficult or time consuming to go back and find the original submission/comments. Whereas if they are already open or within the browser history stack, things are much easier to track.
