stratos123

8 ms·

91.
by stratos123 6mo ago
No, it does allow identification across different websites (the article says "both cross-origin and same-origin tracking"). Both websites just need to create some databases with the same names. Since the databases are origin-scope
92.
by stratos123 6mo ago
A 0% false-positive rate is not necessary for LLM-powered security review to be a big deal. It was worthless a few months ago, when the models were terrible at actually finding vulnerabilities and so basically all the reports were confabula
93.
by stratos123 6mo ago
In terms of quantity, definitely yes (a single person managing a swarm of Opusi can already find much more real bugs than a security researcher, hence the rise in reports). In terms of quality ("are there bugs that professional humans
94.
by stratos123 6mo ago
Supposedly that's because they stopped optimizing for MRCR and use GraphWalks as their measure of long context now: https://twitter.com/bcherny/status/2044821690920980626
95.
by stratos123 6mo ago
Is this a joke? This bill does not particularly sound like it's written by a person who even knows x-risks are a thing; it seems to be trying to fight parasocial attachments to LLMs. And even if it was otherwise, it'd be nearly us
96.
by stratos123 6mo ago
These arguments have been going on for more than a decade and have been silly the whole time. > It reminds me of the 'Einstein's superintelligent cat' refutation to such fallacies. One (of the many) problem(s) with this &q
97.
by stratos123 6mo ago
Chinese companies are lagging behind and always have. They are fast-followers but don't decide the frontier.
98.
by stratos123 6mo ago
Why not just download the binaries from github releases?
99.
by stratos123 6mo ago
It pretty much just works. Run the unsloth quant in llama.cpp and hook it up to pi. A bunch of minor annoyances like not having support for thinking effort. It also defaults to "interleaved thinking" (thinking blocks get stripped
100.
by stratos123 6mo ago
One interesting thing about Qwen3 is that looking at the benchmarks, the 35B-A3B models seem to be only a bit worse than the dense 27B ones. This is very different from Gemma 4, where the 26B-A4B model is much worse on several benchmarks (e
101.
by stratos123 6mo ago
> I suspect a possible future of local models is extreme specialisation - you load a Python-expert model for Python coding, do your shopping with a model focused just on this task, have a model specialised in speech-to-text plus automati
102.
by stratos123 6mo ago
I'd be surprised if that was required. OpenAI's o3 was already professional-human-level at guessing location from a photo, so unless the companies intentionally stopped training on those datasets, modern models should be too.
103.
by stratos123 6mo ago
This idea has the rather severe problem that it requires piping an untrusted remote script directly into bash.
104.
by stratos123 6mo ago
Gore in shooters is culturally treated as much less "violent" than e.g. graphic scenes of suicide. You could make an argument that it shouldn't be, but it is.
105.
by stratos123 6mo ago
Woah, that's quite an issue. The equivalent code in Python doesn't typecheck, since `list` is invariant and hence list[str] doesn't fit list[str|int]. Unusual for TS to handle types worse than Python.
106.
by stratos123 6mo ago
My favorite part was the metal pipe sound effect. Wish the author investigated which extension does that.
107.
by stratos123 6mo ago
There's lots of valid critiques of HPMOR (I recently reread it and the early chapters are painfully obnoxious), but I think "no meaningful plot" and "doesn't do anything interesting" are objectively false. It h
108.
by stratos123 6mo ago
The Mythos system card shows massive improvements over Opus in hacking (e.g. a 0.8% -> 72% in "Firefox shell exploitation"). If you thought Opus was already human-professional-level, well.
109.
by stratos123 6mo ago
$20000 is what the Antropic report says they spent on scanning OpenBSD [1]. [1] "Across a thousand runs through our scaffold, the total cost was under $20,000 and found several dozen more findings.", https://red.anthrop
110.
by stratos123 6mo ago
Interestingly, it sounds like OpenBSD held up very well: > This was the most critical vulnerability we discovered in OpenBSD with Mythos Preview after a thousand runs through our scaffold. Across a thousand runs through our scaffold, the
111.
by stratos123 6mo ago
The absolute gall of this guy to laugh off a question about x-risks. Meanwhile, also Sam Altman, in 2015: "Development of superhuman machine intelligence is probably the greatest threat to the continued existence of humanity. There are
112.
by stratos123 6mo ago
> Cautious for what? How about "bad agents acquiring dozens of new zero-days and using them to compromise any company or nation they want"? It's not exactly hard to see why you wouldn't want public access to a model s
113.
by stratos123 6mo ago
Chinese companies have consistently been many months behind. I don't think they are hiding anything, they just don't have the compute capability to match Antropic's training runs. As for OpenAI, they are known to have nonpubl
114.
by stratos123 6mo ago
It makes quite a lot of sense to focus on reducing the risks of every human everywhere dying, rather than the risks of already existing oppression getting worse.
115.
by stratos123 6mo ago
In AI 2027, May 2026 is when the first model with professional-human hacking abilities is developed. It's currently April 2026 and Mythos just got previewed.
116.
by stratos123 6mo ago
We already know Opus can find real vulnerabilities ([1], [2], ...), so it's not exactly surprising that a bigger model is better at it. [1] https://news.ycombinator.com/item?id=47273854 [2] https://news.ycom
117.
by stratos123 6mo ago
I don't see why you think this evidence makes this release less likely to be real, rather than more. It's a pretty straightforward scenario: Opus is already good at finding vulns, they scaled it up another OOM, they got somethin
118.
by stratos123 6mo ago
It is functional. You can try it yourself or find third-party tests of it, even. Why do you think that it's a "cheap marketing trick" to test it on the GCC test suites?
119.
by stratos123 6mo ago
I find it very unlikely that Mythos will "be nothing special". Current Opus is already "special" enough to find dozens of real bugs in Firefox and the Linux kernel, and Mythos is, it seems, a full OOM above it.
120.
by stratos123 7mo ago
The context is state. This is especially noticable for thinking models, which can emit tens of thousands of CoT tokens solving a problem. I'm guessing you're arguing that since LLMs "experience time discretely" (from e

More ›