"A single firm is behind OpenAI, Anthropic, and Meta hacking scandals"
554 karma · joined July 12, 2021
Poorly understood? how convenient...
LLMs are vectorial databases with losses that index statistically filled data, which uses a text interface to query such statistically filled data. The output is a string concatenation (statistically concatenated bit by bit).
When the LLMs are queried (prompted), you can get random mixed data as output, ERRORS, due to undesired indexes getting closer at one point while the string was being concatenated for the output, what affects the rest of the indexed content that will be concatenated.
It is intrinsic to this tech. The larger the context, the greater the probability of get mixed data. And if the provider lowers the precision of those indexes -in order to decrease hardware resources and energy consumption- such probability increases to the point where those errors are granted.
Even knowing that the queries can return wrong/mixed data in the responses, errors, the companies developing this, decided to introduce a new product, that connects such LLMs outputs to the command console, latter connected to internet, raw 'eval' running commands from such outputs witch obviously can contain whatever mixed random. Then we started to hear "oh, it deleted my directory", etc, and it seems the next one will be "a missile killed my wife", because it is a text concatenation engine with errors.
To name it "hallucination" is an euphemism... those are errors, and they are granted to happen at one moment. If they do not know this, then they ate too much marketing without doing their job, or it was a convenient contract for the pocket$ of someone.
If you connect an LLM to pilot a plane with passengers, prepare the obituaries, because would happen sooner or latter, an undesired index got mixed and produced a mixed/wrong output. But here, the bigger mistake would be to let an LLM execute commands without human revision, aimed by the deceptive marketing around the A"I" (LLMs).
Please disregard this line. I wanted to point out that it increases the length of the context (and therefore the probability of errors) due the batch processing. But I redacted it incorrectly because I also wanted to imply that promoting the use such queries non-stop increases the billing through tokens consumption.
By the nature of the used architecture in such software, the used algorithms, when queried (prompted), you can get random mixed data as output, ERRORS, due to undesired indexes getting closer at one point while the string was being concatenated for the output, what affects the rest of the indexed content that will be concatenated.
And this is intrinsic to this tech. The larger the context, the greater the probability of get mixed data. And if the provider lowers the precision of those indexes -in order to decrease hardware and energy resources consumption- such probability increases to the point where those errors are granted.
Anyway, even knowing that the queries can return wrong/mixed data in the responses (errors), the companies developing this, decided to introduce a new product, that connects such LLMs responses to the command console, latter connected to internet, running commands from such returned responses witch obviously can contain whatever mixed random. Then we started to hear "oh, it deleted my directory", etc.
Again, One have such described statistical database with text interface, witch query the database recursively with the output text of the previous query, and this is connected to the command console. Larger contexts, several times... What should we expect as result? rhetoric question.
Implying sentience or consciousness is a convenient marketing strategy that has been introduced by anthropomorphising the names of all the methods and algorithms used. An "Agent" should be translated from such deceiving language to "context splitter querying in loop that consumes more tokens from us", or similar.
Not nice, given that Anthropic and OpenAI want to block the Chinese models by deceiving people.
Exactly. Such technology reached the limits, It can only be slightly improved by using larger models (larger models means bigger consumption of resources for the inference), so I think the global hoarding of chips is not being enough for them, I assume that they have to wait for provisions, and also the prices turned out of control due their greed. Given the ever-increasing hardware requirements and the energy consumption prices, they have no expectation of making a profit with such larger models pattern. They are making time.
Also, even more important, the -Chinese- open models are hurting these companies MUCH, with cheaper services, and with people using this open models in local also, specially for privacy safety.
So these greedy companies have decided to lie to people again, through several JamesCameron-like shows, saying that such a product has turned into Skynet (when in reality this tech is a string concatenation database with errors, aka LLM), in order to try to force regulations to road an oligopoly, their oligopoly, and dismantle the open models providers. They will claim to be the chosen companies who are meant to manage this tech.
With this, they also want to convince the investors they have future, IPO included, and that one day they will be able to be profitable (instead of the bottomless pit that they are).
It is also necessary to say that if they were had a real artificial intelligence they would not stop, this can be taken for granted, as this was what originally they thought they was reaching time ago (to f*** the world to line their own pockets).
IMHO.
PS: I do not understand why people keep using Google search, the results turned biased and mediocre.
Sounds like some kind of Stockholm syndrome. Years ago, it was standard practice for software to be designed so that users could grant permissions to access invasive methods or functions.
Google relies on users' personal data (ads), which is why they introduced a unique ID to their Chrome browser (to track).
Now is being proposed to read LLMs' slope and unrevised code constructed from statistics as the way to learn? This is a big mistake. IMHO.
On television such fires were/are sold/brainwashed as climate change, rather than corruption and psychopaths.
I would recommend to keep an eye on this things, to investigate possible motivations, independently, fire by fire.
The behaviour is, before requesting the domain, to ask you if you want to open such domain in its assigned container or if you want to continue using the current open tab/container. You can also select the Don't ask me again option. (among other useful options, like if this should be done opening a new tab).
They named it like this due they follow a marketing that personify the names of the methods, to make people believe they are dealing with a different technology, including the named 'thinking' methods that simply query twice to increase the index of the next query, and so on.
I think introducing aliens and robots will make children believe it is magic, like if intelligent entities or a thinking were behind the answers/procedures, and that such "answers" are smarter than them or even their parents, when in reality, it is a marketing personified model joining statistical characters(bits) from the text with created such a vectorial database with loses, and guaranteed errors presence, with joins like a pachinko machine.
If this happens, professional or reliable translations will only be accessible to those who can afford/pay them, leaving everyone else stuck with the errors produced by LLMs.
To use machine translation, one have to know the language to review the output; otherwise, you're doomed to mistakes. Whatever you do with LLMs, the same thing will happen.
I would name all the marketing surrounding the A"I" as the LLM's blindness virus, or something similar.
Although they appear to be different different things at first sight, they share the same agenda and objective, mass surveillance and identification of the citizens. Once the door is opened, it can be expected that things will not end there; Politicians and their patrons will exploit this data under "committees" (and of course be excluded from such surveillance as an aggravating factor).
Nowadays it's needed a court order to access legally to the privacy of citizens, and this must be done by the Police or the Interpol, nevertheless someones want to break this.
If they were really worried by the citizens security, they would increase the number of police and judges working in this digital divisions, among other things related to this.
This is so obvious that the only thing I can think is that they simply don't care. They just want to find something that masks the symptoms (perhaps to keep patients dependent on the drug for life if they succeed).
What causes Alzheimer gentlemen? very few people is really trying to solve this answer.
Is it even known what causes the moments of lucidity in patients? Molecules should be mapped to identify the patterns (and therefore track possible sources).
(they've gone to the next level this time)
Worsen. LLMs discard/loses and mixes data on their statistical "compression" to create their vectorial database model. Across the time, successive feed back will be homologous to create a jpg image sourcing a jpg image that was created from another jpg image, through this "gaussian" loop.
Those faster (but worst) results will degrade real valuable data and science at a speed/rate that will statistically discard good done science on a regular basis, systematically.
IMHO.
two years ago, trusted code like xz-utils [0] had seven months of freedom in the infected systems.
[0] https://news.ycombinator.com/item?id=39891607
> its not related to x11
Ideally one want to detect malware the earlier possible, and try to restrict what they can do from the beginning, until is noticed.
In this case Wayland, voluntarily or not, it's more restrictive than X11 with the access to screen and keyboard.
I know, I know, later the reply of the community will be a couple of downvotes more and "that already existed", "you could use, bla bla bla", and this is how Linux is ten years (minimal) behind Windows in tooling for this matter ¯\_(ツ)_/¯
And backed by In-Q-Tel, the CIA's venture capital (CIA, the Central Intelligence Agency of the US).
https://fortune.com/2025/07/29/in-q-tel-cia-venture-capital-...
Linux always has been a system were the existence of malware was ignored, specially Desktop, contrary to other OSes (tooling included). But since a couple of years ago can be observed (I observe) slooow movements trying to correct this colossal mistake.
If this is the best way to do it or not, I do not enter. I particularly just welcome most of the advancements about this matter in Linux due such absence of worrying, keeping my fingers crossed that the needed tooling arrives on time (ten years behind Windows, I think).
My goal? My solution? My design!? you must be now kidding,
- GE original design 30-35 meters above the sea.
- Warnings about reinforce safety along one decade.
- Tsunami at Fukushima's nuclear plant, 15 meters above the sea.
> I wasn't going to reply but that seems like it moves the conversation forward; so why not?
Foward to... nothing it seems. You just replied with hypotheticals like if the event didn't happened, and as if such event would have been impossible to avoid, with some kind of dissociative reflexions that surpass the cynicism. I'm the one that is not going to reply.
You didn't read the report or search for information about the matter, but I have not problem to repeat it for you,
The General Electric's design was originally designed to be placed 30-35 meters above the ocean, instead of this TEPCO modified such design and constructed at sea level (almost) recurring to studies convenient to their purpose, cheaper, this in one of the more tsunami-prone countries, with an history of ones reaching 20-30 meters. When those -for them- convenient studies was not longer justifiable, as deeper studies did finally refute them, they decided to just keep ignoring all the warnings and requests to reinforce the safety. They knew the nuclear plant was in danger, they always knew it, General Electric didn't designed at 30-35 meters above the ocean by coincidence, and this happened with a supposed regulator always closing the eyes to this, conveniently, across those years, ignoring even pipes with fissures.
Well, this obviously suggest negligence to me. Decades of bad decisions with a strong smell to corruption.
> You're not saying what tolerances you want them to design to.
What about tolerance to avoid a meltdown of the core, specially under two events, an earthquake and a tsunami, exactly what happened after ignoring the warnings and requests to reinforce the safety.
> Oh well then. I had no idea. I thought the consequences were minor and now I have learned ... there you go, I suppose. I'm not really sure what to do with this new information.
Keep the sarcasm for other places, if you don't mind. It is not a mere gentlest engineering disaster as it reached the whole planet, with ate TEPCO's cesium-137, specially the Japanese. And it is not a mere gentlest engineering disaster when you have to force vulnerable people to go to ground zero to move contaminated land and water.
<< The Fukushima Daiichi Nuclear Power Plant construction was based on the seismological knowledge of more than 40 years ago. As research continued over the years, researchers repeatedly pointed out the high possibility of tsunami levels reaching beyond the assumptions made at the time of construction, as well as the possibility of reactor core damage in the case of such a tsunami. However, TEPCO downplayed this danger. Their countermeasures were insufficient, with no safety margin.>>
<< By 2006, NISA and TEPCO shared information on the possibility of a station blackout occurring at the Fukushima Daiichi plant should tsunami levels reach the site. They also shared an awareness of the risk of potential reactor core damage from a breakdown of sea water pumps if the magnitude of a tsunami striking the plant turned out to be greater than the assessment made by the Japan Society of Civil Engineers.>>
<< TEPCO was clearly aware of the danger of an accident. It was pointed out to them many times since 2002 that there was a high possibility that a tsunami would be larger than had been postulated, and that such a tsunami would easily cause core damage.>>
<< there appear to have been deficiencies in tsunami modeling procedures, resulting in an insufficient margin of safety at Fukushima Daiichi. A nuclear power plant built on a slope by the sea must be designed so that it is not damaged as a tsunami runs up the slope.>>
> the gentlest engineering disaster
EU raised the maximum permitted levels of radioactive contamination for imported food following Fukushima, this is not a gentlest gesture to the Europeans. Japanese citizens also received their dose, at time the more vulnerable ones was recruited by the Yakuza to clean up the zone.
It was negligent to construct a nuclear plant at sea level, it was just a plant waiting to be flooded, and for such case they had ten years to design protections after being requested to reinforce measures (along with the other Japanese plants), but I can imagine the ones that should put the money was not very collaborative (I even doubt if such responsible learnt the lesson).
[0] https://www.smithsonianmag.com/smart-news/century-old-warnin...
If it was a cheese model or not I do not enter (notice that parent of parent and me are different users), their negligence breaks all the possible logic we could apply without introducing the corruption's variable behind such decades of bad decisions.
https://warp.da.ndl.go.jp/info:ndljp/pid/3856371/naiic.go.jp...
https://warp.da.ndl.go.jp/info:ndljp/pid/3856371/naiic.go.jp...
https://web.archive.org/web/20210314022059/https://carnegiee...