Well, of course there is "Selbstreparaturparadoxon", but nobody uses that. This is more like a Münchhausen-Problem, where Baron Münchhausen pulls himself out of the swamp by his own hair.
It's also a lot easier to get caught when you're not paying full attention to whatever is being sent your way. Say an algorythm starts playing AI music while it's just playing in the background, or a friend posts something silly that you read sideways.
I mean I've seen people saying that whole "oh, it's just so good you can't detect it" a lot but not any actual evidence. I guess sentence can be short enough to not look clauded, but aside from that smells are still very common even for the best frontier models, so how anything else would be indistinguishable ?
I would argue that Krebs is exactly the type of guy who will do anything for a good story, whatever it costs. I saw him lying more than once and this eroded my trust completely.
Also frenck explained how the feature works and why this is save.
On the other side Viss basically just said "trust me bro, it is not save and by the way, I'm THE expert and I'm so very smart, that I don't even need to list some facts why this could be unsafe, just fix it NOW", which I find exhausting, unhelpful and boring.
Thats the thing though, it's not safe [1] and puts trust in a third party - HIBP.
To do that automatically without any notice to the user breaks the fundamental value proposition of a local home assistant which over and over again preaches keeping things local and private.
As someone who relies on home assistant every day, I'm quite disappointed to hear Frenck's response to this.
From the issue author: "I can give you a list as long as my arm about how I could abuse this to loot credentials from every HA install on the planet.".
From the linked article I can count 2 with following conditions:
1. relies on a malicious or compromised lookup server, linking lookups to the same user, the user checking two predictably similar passwords, both passwords being breached, the attacker having their plaintext candidates, and knowing the account identifier for account takeover
2. relies on a malicious or compromised lookup server, linking lookups to the same user, the user checking multiple predictably related passwords, sufficient computing resources to search the original password space, and knowing the account identifier for account takeover, the attack demoed seems to work if you have a weak password to starts with (10 chars alphanum).
Yeah, I think the situation is that if your password is already weak/known to a level that it should already be considered compromised, a malicious HIPB could surface that compromise a lot more reliably/quickly.
Because it is happening without user's consent. Imagine you invite me for a drink in your home, and while you're not looking I go and open the front door, for whichever reason. How would that make you feel?
This looks like two different problems to me: one is the expectation of the behavior of the service (whether it talks with an external service at all) and another one is if the service is technically safe (the call to the external service makes my passwords guessable or not).
I can't argue with your analogy, because it doesn't make sense to me.
The tool looks interesting, but man it is sooo hard to read the LLM generated blog post as a human. Not because the text is bad (it kind of is), but because it doesn't feel like anyone has put any energy and real work into it. So it feels like a waste of my time to read something, that could have been three short sentences and a benchmark table without loosing much information.
I thought the blog was pretty reasonable until I came across "One caveat keeps the speed numbers honest" which was the one obvious tell for me. Otherwise I thought it was pretty straightforward to read.
Curious, what were the tells. I'm mostly familiar with Claude family of models' writing so I wonder is this Astra or similar which is why I didnt find it odd.
To play devils advocate: Isn't it that the management had to choose between two potential problematic decisions? What if the management declined the hindu request and then the hindus made a big fuss about it? Wouldn't people scream "how could they think that discriminating the hindu religion would be a good decision?!?".
France has a storied history of resisting efforts to impose non-western, extremist beliefs on them
You could start with Charlie Hebdo for an idea of how it might be received if a religious extremist group wanted to fuss over hiding French women from the public.
