Investment isn’t debt, but investment and debt demand revenue in functionally identical ways, and OpenAI has taken an enormous amount of investment, so they are subject to broadly the same pressures as a hypothetical company that had borrowed lots, no?
> investment and debt demand revenue in functionally identical ways
No they don't.
A company has legal obligation to pay loans at the agreed rates.
There is legal obligation that defines rates of return beyond "acting in shareholder interest" (and OpenAI is a public benefit corporation so even this is reduced).
Does it bother you that the drumline consists of two bars recorded on a synthesizer and then repeated over and over again with inhuman, flawless precision for the entire duration of the song?
Recorded on a synthesizer/loop machine by a person imperfectly hitting the pads, then repeated while two humans imperfectly sing and play instruments with the backing drum loop. It's not the same thing at all.
A lot of electronic music is also someone sitting on a loop machine and "playing" their samples.
IMO it is a little bit silly to describe “a system that works well until it intersects with a trillion dollar company’s PR campaign” as an outdated failure.
Or perhaps, trillion dollar companies should be more responsible. Either because they fear regulation that will force them to be responsible, or because there is regulation that will make the responsible.
I agree. At present I do not see any set of incentives, meta-incentives, etc., that are extant, or even rigorously imaginable[0], that will lead to such a state of affairs.
0: credit to Xenosystems for this excellent phrase
I mean Patch the Planet. This was the first I had heard of it, and I assumed from seeing the press releases from OAI working their way through journalism’s digestive tract that it was announced today - but perhaps it was previously announced.
If that is case, I’ll refine my statement to “Also happening today: OpenAI is doing a PR push on their software vulnerability initiative”, which I think suffices to convey the same meaning as the original.
I’d like to nominate the graph there for some sort of crime, for using a log scale from 1e-55 to 1, where there is one result at 1e-55, one result at 1e-18, a million results at 1e-3, the omgwtfbbq threshold drawn at 1… and each result is drawn as a bubble that covers 3 orders of magnitude above and below.
Except because it's some sort of a vibecoded atrocity, nothing happens if you try to toggle from "all" to "last 3 days", it only works if you change the value (e.g., to 2 or 4 days).
Stuff like that is really frustrating because it's not like we didn't have one zillion decent, human-tested graphing frameworks before.
I think it's basically an alignment problem. LLMs tend to think it's better to just do something small and bespoke than to just pull a mature library and use it correctly, which, to be clear, they are perfectly capable of doing.
And this is in part a reflection of our own bad preferences. This is what always annoyed me most about the left-pad Discourse a decade (oh god, a Decade™) ago; "why do you need a library for something so simple in the first place!?" people would scoff. And of course the answer is that even the reference class of left-pad-like problems includes a bunch of examples where the problem has bizarre and sticky edge cases you'd never consider, so in general, "just pop in a popular and well tested library" is actually a pretty good policy (assuming your package repository doesn't allow anything really silly like deleting entire packages and breaking everything downstream).
Of course charting is more obviously complicated, but again, it might seem fairly simple if all you're thinking about is one kind of chart in one style with limited settings. The difference is that for a human coder, "just whip up some chart code" is enough of a pain in the ass that they'll reach for a library by default, whereas a coding agent can just spit something out which mostly works in a few seconds.
All else equal, I'd rather have the richest possible functionality built into the language. At least I know that it will be updated and maintained in one place, and there's very little ambiguity about setup. You don't have to start picking apart the fact IN THIS PARTICULAR PROJECT, libfoo was imported by pasting it in the code base in 1977 before package management existed and globally namespaced, but libbar is a first-class citizen in the depency-injection system.
If you know you can just call bargraph($data), even if the stock bargraph library is boring, then people and LLMs will favour it.
I suspect they used an underpowered model, or did it a while back. I would be surprised to see 6.1 Sol or Astra produce those bugs, even one-shotting. Instead, I'd just expect it to have broader product issues rather than out and out bugs.
Eh, then it just shows relative change between results; in isolation, yes, not contextualising these results at all is better than misrepresenting them, but doing both is arguably worse than either, it kind of admits you know exactly what to avoid saying.
So from what I gather, it was found by OpenAI (via unspecified model?) and independently by Wiz (via Gemini Argon)… but before that it was found by Calif (via unspecified but consumer-available model)… but Calif credit LCFR_HAX before them with first discovery, disclosure, and remediation efforts.
I guess either LCFR or Calif put it in the training data (by using a consumer-available frontier model in the process of discovery or remediation), and then OpenAI reproduced it from the training data, Navier-Stokes-scoop-style? Except I guess more plausibly unaware, and also late? They thought they had something novel, because they weren’t privy to the existing remediation effort? It’s all rather confusing, and then on top there’s a leak and a PR campaign and it just becomes impossible to untangle.
I don't think the timelines make sense for it to have been included in training data, LLM training does not seem to be nearly efficient enough to learn these things from lone examples, and it wouldn't make sense for this earlier work to have made its way into both OpenAI and Gemini's data. Navier-Stokes is not strong evidence that this kind of thing can happen, as OpenAI denies that Buckmaster's work was in the training data for the model they used, Buckmaster didn't have a blowup that they could steal, and Scientific American reports[1] that the similarities between their approaches were superficial. It's also well-established that models can find novel vulns at this point.
My understanding at the time I wrote that was “someone else using chat-logged OAI model: September???; OAI using unspecified model: October???”, and when I say “Navier-Stokes-style-scoop” I am thinking along the same lines as the comment I made during that controversy, here: https://news.ycombinator.com/item?id=49639648. I don’t buy that their public statements are not “technically correct but misleading without internal context” on these matters.
Naturally, with the added info from the PHP foundation link (specifically that OAI reported it the same day as Calif, and that Calif was operating under the auspices of an OAI program) the timeline collapses for my initial impression!
> On August 11th researcher Ryan A at Calif reported an unconditional remote code execution with the same root cause as the OOM problem we had found in 2024. OpenAI independently reported the same root cause on the same day
> The PHP Foundation got the PHP project into the Patch the Planet program, as part of which security firm Calif began an audit of PHP on July 30th, 2026. On August 11th researcher Ryan A at Calif reported an unconditional remote code execution with the same root cause as the OOM problem we had found in 2024. OpenAI independently reported the same root cause on the same day
How does this work? Calif works with OAI on Patch the Planet program. Calif reports the finding to PHP. OAI ‘independently’ reports it too, on the same day. I suppose Calif did communicate with OAI about said findings, or is this not how Patch the Planet operates?
It seems extremely coincidental.
Could that “single instance” be copied enough times in places also used in training, so represented much more in the training data than once? If it only takes a small number of malicious chunks of information can potentially poison an LLM or large size (ref: https://www.turing.ac.uk/blog/llms-may-be-more-vulnerable-da...), then presumably a small number of a good chunk can influence output similarly?
I can literally feel my eyes physically defocusing trying to read this, and I can’t help but conclude that, at a level below conscious thought, my brain is trying to defend itself from AI-generated text. Perhaps it is more of a cognitopoison than I explicitly knew.
Ctrl-Alt-R for reader view in Firefox, not that it makes the whole thing much more interesting to look at - we'll forget about it all soon enough anyway. Don't forget to tweak the default settings for reader view for a better experience.
reply