TeMPOraL

17 ms·

391.
by TeMPOraL 3mo ago
No, it's not. That worked for GPT-3 level models, all further models were trained to ignore it. And yes, a small child would jump off the bridge if an adult told them to. Hell, urban legend says Harry Potter books managed to convince a
392.
by TeMPOraL 3mo ago
> The same could be done in an accelerated manner to LLMs with some kind of DNA-like mechanism related to weights. It is plausible that LLMs will be bred in the future for specific roles by how well they fit - kind of like continuous pa
393.
by TeMPOraL 3mo ago
> You can spoof your caller ID to make your scam more convincing. I recently came closer than I'd like to falling for a scam (read: I picked up a call and conversed with the caller for 30 seconds before realizing it's a bot),
394.
by TeMPOraL 3mo ago
> note how often Opus 5 now says things like "task X was blocked by the classifier, I will not attempt to circumvent that" Interesting. I had an issue with Opus 4.7 / 4.8, where it would sometimes flake out on a task, an
395.
by TeMPOraL 3mo ago
> Base64, Unicode substitution, emojis, output of obfuscated but "harmless" code run in a sandbox, image steganography, etc that could be endlessly disguised without a human even being able to see it, yet alone fall for it Li
396.
by TeMPOraL 3mo ago
Sometimes it's the only thing you have available. Like IDK during a fire in a basement server room, where the only connected device available is a laptop with wired connection and an open inbox. Because you know, you tried IM but &quot
397.
by TeMPOraL 3mo ago
Even financially, Hugging Face benefits directly from any and all interest in AI.
398.
by TeMPOraL 3mo ago
The elephant in the room is still that this was done by an LLM and not by a human . Script kiddie or not, in this reality, in the year of our lord 2026, people still call SOTA LLMs "useless", "scams", and "stocha
399.
by TeMPOraL 3mo ago
You're saying all this is a distraction, basically giving the forensics researchers enough exciting material to make them conclude their job is done, while the actually intended attack remains undiscovered?
400.
by TeMPOraL 3mo ago
Intermittent fasting? Alternative medicine? Fitness? All the beauty press and anorexia epidemic in adolescents? Fashion model industry? Smoking ? Political propaganda inventing to broadly-understood terrorism? Not everyone falls for any of
401.
by TeMPOraL 3mo ago
Haven't seen that article before, thanks for the link! Having read it now, yes, it's arguing roughly the same point as I am. I say roughly because e.g.: > Notice that all these strengths involve transformation, interpretation,
402.
by TeMPOraL 3mo ago
Different layers of abstraction. You can look at it this way: the machine separating instructions and data can still emulate a machine that doesn't. Within the inner machine, there is no such separation. Outside of it, but still within
403.
by TeMPOraL 3mo ago
Whatever was released in the open as weights is forever out there and beyond reach of any corporate interest.
404.
by TeMPOraL 3mo ago
Which can be unsegmented and untagged with a soldering iron, electron gun, and/or firmware patch, depending on how it's implemented.
405.
by TeMPOraL 3mo ago
You control tokenization. But the system able to tell you what those tokens means is the very one you're feeding the tokens to.
406.
by TeMPOraL 3mo ago
Is there any major for-profit AI company that doesn't have generous free tier with access to SOTA models?
407.
by TeMPOraL 3mo ago
Because you can access it for free? As it was the case from the first day LLMs became a thing in public consciousness? Or did I miss the change, and ChatGPT, Claude and Gemini no longer have a free tier anymore? And then the next tier that
408.
by TeMPOraL 3mo ago
Foldables may be an exception (or at least that's what I'm telling myself). The combination of larger screen when unfolded + logistics of a phone when folded means a foldable is basically a tablet that you don't have to activ
409.
by TeMPOraL 3mo ago
The market must be stalling to the point the companies can't afford to innovate anymore. Or at least that's how I understand why Samsung recently introduced literal first-party program for renting phones and is advertising it heav
410.
by TeMPOraL 3mo ago
I was thinking the next generation of attacks will involve playing the system prompt & harness the same way return-oriented programming does with normal binary code. That is, setting things up so that the system prompt itself implies th
411.
by TeMPOraL 3mo ago
Any way to do that AND use subscription instead of per-token pricing from SOTA providers?
412.
by TeMPOraL 3mo ago
You're saying that people fall for phishing because scammers invent completely unrealistic scenarios that would never happen outside TV shows?
413.
by TeMPOraL 3mo ago
What invent human consciousness? > LLMs are vulnerable to classes of attacks that humans just aren’t Name three that don't have direct analogues with humans.
414.
by TeMPOraL 3mo ago
Any good AI will just react like in https://xkcd.com/1494/ . This is an example of where the lack of "instruction/data" separation is a benefit - the system is able to recognize you're obviously tr
415.
by TeMPOraL 3mo ago
The mistake is in treating the LLM as just another deterministic, narrow computer program it can reason about. It's not. It's a "DWIM" system, and unless you can always precisely express what you mean - which you can&#x2
416.
by TeMPOraL 3mo ago
Right. But even with a DBA, the possibility remains . We accept that. That's kind of my point with fighting against the "lethal trifecta" and "code vs data" mindset - once people engage cybersecurity mindset, they&
417.
by TeMPOraL 3mo ago
> A pure Harvard architecture machine has exactly that separation. It emulates and enforces that separation. A mathematical abstraction of a Harvard architecture machine has that separation, the real machine merely emulates it, and
418.
by TeMPOraL 3mo ago
To drive the point about this being fundamentally unsolvable home, imagine a variant of this scenario. I could write an email that is directed towards that person, that says WE ARE STUCK IN THE SERVER ROOM AND THERE IS FIRE STARTING. PLEASE
419.
by TeMPOraL 3mo ago
In some sense you're firing a human and hiring a new one each time you start a new conversation / clear the context window. My point is at the systems design level. LLMs as components are a substitute for people, not regular softw
420.
by TeMPOraL 3mo ago
> (unless there is some very specific context). That's the key thing. That's why you neither can nor want to introduce any kind of code/data separation into LLMs. > With AI systems, we have the benefit that the disti

More ›