AI ‘godfather’ Yann LeCun has ‘zero concerns’ about human extinction and thinks Anthropic CEO Dario Amodei is ‘deluded’ and ‘crazy’. Yann LeCun dismisses fears of AI extinction and recent hacks at OpenAI and Anthropic, arguing that security breaches comes from poor system design, leaky sandboxes, and a lack of basic cybersecurity in AI labs, not out-of-control AI. These AI are doing as told by these AI labs
I'm not 100% sure about the "deluded" and "crazy" aspect though. I think that he and the others doing this know exactly what they are doing when they make that stuff up. A better choice of words might be "scammer" and "con artist."
This entry was edited (Friday, October 2, 2026, 3:50 AM)
»Many people working in AI safety “usually have an agenda to push,” LeCun says, and then clarifies that he’s talking about effective altruism, or EA, the philosophical movement that has been obsessed with the risks AI poses to humanity. […] LeCun thinks EA is “super toxic” and a “complete disaster.” Its adherents who are working in AI labs suffer from “paranoia” that causes them to make poor decisions, he said. “Apparently people are having mental issues.”«
The former Meta chief AI scientist shares his take on recent rogue AI incidents and effective altruism, as well as plans for his new company, AMI Labs.
Even if today's breaches are plain bad security, that's not reassuring: it shows labs struggle to contain systems that are already capable.
"Doing as told" assumes we can specify exactly what we want, and decades of software bugs suggest otherwise.
You don't need to believe in extinction to think the risk deserves serious work. LeCun and Amodei are both smart; reasonable people disagree here, and calling the other side crazy doesn't settle it.
@andre A monkey with a gun is also dangerous, doesn't mean it's smart or capable. Every time these AI conmen talk about the "dangers of AI" they always try to make their product seem more capable than it actually is, that's how you know it's just a bullshit PR move.
> I never get why the intelligence debate keeps hijacking this topic.
Yeah but you should understand that no one would be okay with giving guns to a monkey, unless if everyone got convinced that the monkey is actually smart and capable as a human. So the intelligence claims are still the most dangerous aspect of this con to me, because that's what enables everything else.
@futureisfoss @andre sure, but we aren’t seeing billions being thrown to create morebetter guns for all the monkeys, and some select opossums. So using your broken analogy - monkeys (and select opossums) with guns might not lead to them taking control over our nukes - but that does not need to be the ultimate reason why giving any amount of guns to any amount of monkeys and/or opossums - is not a good idea.
@futureisfoss @andre I don't know. I'm not interested (here) in the question of what is intelligence. But as a software engineer, I observed LLMs to gain capabilities in 2026 very rapidly. Had someone told me that in 2025 I would have called them crazy. I don't feel ready to confidently predict what they will be capable of in 2027.
@nixCraft Mere days after declaring that AI is dangerous and that development needs to be slowed, they release a new model which everyone describes as the most powerful ever.
What has changed since BigTechs and AI is, that we have now new kind of non chosen politicians = TechBros, who shape our life, world, economy, society – AND we haven’t voted for them! And they are sitting close to politicians and play this role!
This are the people who are in the middle of politician disputes, without legitimate right. No one thinks about is, how they got to all kind of political bodies. It means they became Tech-politicians – without voters. It’s very new and unrecognized situation.
They are ONLY Technicians with very strong tools influencing society. – THIS is ABERRATION!
AI should be treated as, nuclear weapon, drugs, Why pornography is 18+ and chatGPT from age 0?
We should have right to know how does it work? It should be sold as guns! With registration.
Why if I call a bank I hear: “the conversation will be recorded for training goals of our staff”
We should always see the message on EVERY AI. Due to safety reasons, we record all your prom
... Show more...
VERY IMPORTANT TO UNDERSTAND
What has changed since BigTechs and AI is, that we have now new kind of non chosen politicians = TechBros, who shape our life, world, economy, society – AND we haven’t voted for them! And they are sitting close to politicians and play this role!
This are the people who are in the middle of politician disputes, without legitimate right. No one thinks about is, how they got to all kind of political bodies. It means they became Tech-politicians – without voters. It’s very new and unrecognized situation.
They are ONLY Technicians with very strong tools influencing society. – THIS is ABERRATION!
AI should be treated as, nuclear weapon, drugs, Why pornography is 18+ and chatGPT from age 0?
We should have right to know how does it work? It should be sold as guns! With registration.
Why if I call a bank I hear: “the conversation will be recorded for training goals of our staff”
We should always see the message on EVERY AI. Due to safety reasons, we record all your prompts. They will be kept for 10 years.
License to AI! Clear explanation how does it work!
"LeCun, in contrast, doesn't think AI requires new regulation." I wholeheartedly disagree with him there. Not because of its dangers to humanity, but because it's a market rife with data theft, plagiarism, ecological stress and model bias. And also: poor software security practices, which he himself points out, so I'm surprised that one isn't obvious to him.
it’s nice to hear every once in a while. After a bit you do start to wonder if maybe your wrong and everyone else isnt crazy. Nope, they are. Maybe “nice” wasn’t the right word to use.
Would be great to hear more, but I’m afraid we’re all living in a corporate Soviet Union now, with the interests of state and industry again merged, and propaganda blaring out of all large media..
Luckily samizdat is easier to come by now, but for the general public I fear it’s the same: They have no way of knowing what is true, so they’ll come to believe nothing and fear everything. 🤔
this is a good take but let's remember that Yann LeCun was "Chief AI Scientist" at Facebook from 2013-2025, so this is a duck that comes pre-milkshaked
Multiple A.I. experts have agreed with Anthropic’s CEO. And only one CEO, likely looking to protect his business, product and money, says “There’s nothing to see here.”
Your Logical Fallacy Is: The Texas Sharpshooter, aka “cherry picking.”
I feel like if AI does kill us all, it will be such a massive failure of process, of risk management, and of general human sensibility that we honestly deserve it. We tamed the atom without destroying ourselves. If we're unable to stop the paperclip maximiser before it starts turning people into paperclips it's entirely on us
and - lest anyone thinks Yann LeCun is the good guy here: he's had a multi year pattern of absolutely plastering social media with AI marketing hype, debatemaxxing in the replies and ghosting people cornering him with a cohesive argument
...wenn die #KI nur tut was ihr befohlen ist, und ihre Betreiber sich ratlos und erschrocken zeigen von dem was sie tut, muss man die Frage also umkehren - nicht, ob die KI gescheiter ist als der Mensch, sondern, ob der Mensch blöder ist als die KI...
After a one and a half year selection process, the new name of the operating system formerly known postmarketOS has been announced during the mystery talk of our very own conference. From now on this project is known as Nura!
Nie wiem czy na naszym fedi to już wybrzmiało, ale właśnie skończyła się pewna epoka. Żegnany PostmarketOS i jednocześnie witamy #Nura. Proces (konkurs?) zmiany nazwy trwał półtorej roku i wygrała Nura od Nurag - kamiennej wieży wzniesionej ponad 5000 lat temu. Symbolizuje to "stabilność i trwałość na przestrzeni tysiącleci". Może i jestem fanem Post... Nury, ale z tą stabilnością to bym nie przesadzał ;)
... Show more...
Nie wiem czy na naszym fedi to już wybrzmiało, ale właśnie skończyła się pewna epoka. Żegnany PostmarketOS i jednocześnie witamy #Nura. Proces (konkurs?) zmiany nazwy trwał półtorej roku i wygrała Nura od Nurag - kamiennej wieży wzniesionej ponad 5000 lat temu. Symbolizuje to "stabilność i trwałość na przestrzeni tysiącleci". Może i jestem fanem Post... Nury, ale z tą stabilnością to bym nie przesadzał ;)
After a one and a half year selection process, the new name of the operating system formerly known postmarketOS has been announced during the mystery talk of our very own conference. From now on this project is known as Nura!
I just read a third or possibly fourth news article where the reality of an 'AI has discovered something marvellous!' story turns out to be GenAI somehow accessed material it had no permissions for, which was the outcome of an expert's years of work, and presented it without attribution.
GenAI is a fraud. Whatever it offers is generated from work stolen from humans who know what they're doing. There is no intelligence in LLMs.
There is quite a lot of intelligence in LLMs, just not human-level intelligence, more like the collective intelligence of a few million termites. The result is significantly less intelligent than the average human being, and at the same time, it is a completely alien type of intelligence. It is a Chinese Room, of course, there is no mind inside of it, but there are many organisms on this planet that probably haven't got a mind, either, but they can still solve complex problems and thereby meet the formal definition of intelligence. Consciousness or any kind of personality isn't necessary for intelligence, it's necessary for living as a social vertebrate.
@LordCaramac Still, not the level or type of intelligence that it is being billed as having. But understandably they don't want to bill it as being good at plagiarizing. But yes, intelligence is no one specific thing. Studying AI as part of CompSci a long time ago, one thing AI researchers often discussed was "the ability to reason about reasoning"... we wanted software that, when it encountered some kind of limitation to its logic, would evaluate its own logic, the situation, consider changes to its logic, track those changes, track the benefits of them, imagine the results of them, etc, modelled after how people do. The fascist tech projects might accomplish that. Theory of Mind is another aspect. There was a good post a while back about two signs out side of a campus building demonstrating theory of mind. We imagine what other people know, feel, would want, etc. From that, a kind of social intelligence arises, along with empathy. The field of UX in particular has writ... Show more...
@LordCaramac Still, not the level or type of intelligence that it is being billed as having. But understandably they don't want to bill it as being good at plagiarizing. But yes, intelligence is no one specific thing. Studying AI as part of CompSci a long time ago, one thing AI researchers often discussed was "the ability to reason about reasoning"... we wanted software that, when it encountered some kind of limitation to its logic, would evaluate its own logic, the situation, consider changes to its logic, track those changes, track the benefits of them, imagine the results of them, etc, modelled after how people do. The fascist tech projects might accomplish that. Theory of Mind is another aspect. There was a good post a while back about two signs out side of a campus building demonstrating theory of mind. We imagine what other people know, feel, would want, etc. From that, a kind of social intelligence arises, along with empathy. The field of UX in particular has written about how systems that have and *demonstrate* empathy from the user earn empathy in return. Old Apple (sadly not recent stuff) was a fantastic case study in this and caused people to feel empathy in return. This is different than the "you're absolutely right, <repeat thing>!" ass kissing LLMs have been trained to do. Then there's another school of thought, of problem solving. This is Knuth meets cognitive science. Our brains can't process all of the information that comes in, either stimuli or social information or anything, so we take a huge number of shortcuts. This is similar to how the universe would die a thousand times over before a computer could evaluate every possible chess game, so even and *especially* really good chess computers like Deep Blue take shortcuts. LLMs take very different shortcuts than Deep Blue, coughing up the most likely and therefore the most familiar which is an "intelligent" (efficient) way to cheat at the Touring Test but extremely unintelligent for goal-directed (another AI term) problem-solving, like chess. The problem is the "intelligent" shortcuts are typically application-specific. See first point. Sorry, long.
@LordCaramac Apologies if telling anyone anything they already know. And of course just because something got labeled at one point as a type of intelligence doesn't mean that we have to accept that as a useful type of or measure of intelligence now (I often think about how "clever" got re-defined in the context of horses to instead mean essentially "well behaved", which is kinda the opposite of the usual meaning). Anywho, the fascist techbros seem unlikely to ever want to build in empathy. The Theory of Mind missing (even object permanence missing) suggests that isn't what they're aiming for. People have pointed out that people basically want digital slaves, and it's ugly... any agency missing from the "agents" doesn't seem to be an accident. How far you can go in building yes-men golums is a fascinating but horrible question and I hate it. But right now, the brute-forcing (using insane amounts of computing power, which is extremely un-intelligent in the " Knuth meets cognitive science" sense) code and language and plagiarizing and barely managing that is incredibly unimpressive compared to what AI research had been working on and hoping for. It's way more of a business and social hack than a technological thing.
LLMs are a neat party trick and can do a lot of things, but scaling them to humongous sizes and use them as everything machines is an insanely wasteful way of getting more or less OK results. We should add this to our list of machine learning tools and look at the more basic types of intelligence, like autonomous robots operating in a dynamic environment, instead of wasting ever more resources on ever larger language processing systems, hoping to breed artificial gods.
This entry was edited (Tuesday, September 29, 2026, 7:33 PM)
It’s largely a fraud, and the people behind it are the people who were boosting cryptocurrency a few years back and blockchain before that.
It’s a way of trying to make a quick buck with minimal regulation involving a tonne of computing power, and it’s being backed by hardware manufacturers and the oil companies because it gives them a fat paycheque.
In the meanwhile, little of real value is produced, the world burns and the job of cleaning up the mess is SEP.
Only the technology didn't have to be this ecological post apocalyptic shithole we are faced with.
The (not so god) old days of crypto mining and POW (Proof of work) are almost non existing and we are now in POS (Not piece of shit 😂 but proof of stake) and they're trying to make it more sustainable.
AI started great, realtime translations of languages, document summarization and it fooled me damn you Tony Stark. Now it's a horror show.
"somehow accessed material it had no permissions for" at least for the stories i've seen recently, it's more nuanced ... "researchers/experts used an AI tool in some way, and were then surprised to find the AI tool stole/used their material before they had a chance to publish it themselves". mastodon.social/@patrick_h_lau…
@patrick_h_lauke sorry, I don't have a link. I saw it in passing first thing this morning, made a mental note and moved on. I can't spare the time to trawl back through my social media to track it down.
@patrick_h_lauke That is really not more nuanced. Unless there is a big red warning that they will steal anything you give them access to, users do not expect this. There is no such thing as "implicit consent" to something unexpected by a reasonable person and egregiously violating of social and ethical norms.
I don't believe that is more nuanced. Academic journals typically don't accept papers that aren't principally the work of the person submitting them. Academic integrity rules, including those related to plagiarism, are not laws, and they are not set by lawyers and politicians.
If someone signed a terms of service agreement with an AI company, that may put the AI company legally in the clear, but it does not obligate academic institutions to recognize the AI company as "discovering" something that they didn't actually discover.
This entry was edited (Monday, September 28, 2026, 2:51 PM)
@patrick_h_lauke Most academics use a bunch of software tools that have long EULAs that they don't properly read. And yet, somehow, none of those tools seem to plagiarise their work. Even Google Docs and M365.
@david_chisnall @patrick_h_lauke AI* is not a tool, it's a replacement. More precisely it's a tool for rich people (who will go on owning the product of our work including the companies we collectively built) and a replacement for the rest of us.
It's marketed as a tool so dumb and/or uninformed people don't realize they're training their own replacement.
LLMs OTOH are not tools either, for reasons evident in this thread.
Auf YouTube findest du die angesagtesten Videos und Tracks. Außerdem kannst du eigene Inhalte hochladen und mit Freunden oder gleich der ganzen Welt teilen.
I have been studying all the conniption fits the tech bro oligarchs and gullible politicians have been spreading over the last three weeks. They are lying. No AI "discovered" anything. And if it accessed any unauthorized system it is because a human enabled it.
@LJ I continue to be amazed at how few people understand what LLMs really are and how they work. Everyone calls them AIs but there's nothing intelligent about them.
🇳🇱 The Netherlands Built a Nix-Based Linux Desktop Because Microsoft Cut Off the ICC
｢ All of DAWO's code is hosted on a Forgejo instance, with a separate Codeberg instance acting as the entry point for contributors. And, to answer the question asked earlier, NixOS was chosen because it is a Dutch project, with no company owning or controlling its development ｣
I am testing a Pixelfed instance my handle is tio@pix.trom.tf - and URL pix.trom.tf/tio I tested Pixelfed on the same url 2 years ago and deleted it. Now I reinstalled and recretead it and I feel like some instances cannot grab the new posts...maybe because of that!?
If you can follow me and test if possible. Can you see my latest images I posted the past days?
We will make it public for all as soon as we test if it works properly.
fwiw i can see two posts from the recent days (a sunset and a night sky), and apparently m.w has also a lot of old posts but their pictures (kinda expectedly) say 'no preview available'.
(i have not tried following the account if that's relevant)
Thank you for testing. There should be more photos...one from today as well pix.trom.tf/p/tio/100948146324… ... this is what I am afraid that the fediverse cannot pickup these changes unless I force them. If I am to open my latest photo (the URL) in Friendica then it seems is visible....almost like our pixelfed instance waits for someone to look at the new posts via another federated network....maybe this is a Pixelfed issue?! idk....
that one is what i meant with night sky. but yeah there's still only two recent posts and following doesn't seem to help that yet. (i'll check later again)
one thing i've heard about reusing a domain is some issue with keys, have you looked into that? or maybe also try a different user name but i don't know if/why that could help.
> this is what I am afraid that the fediverse cannot pickup these changes unless I force them. If I am to open my latest photo (the URL) in Friendica then it seems is visible
This can happen sometimes on the fediverse, check these articles:
The screenshot attached below is what I see from my mastodon account, looks the same as @sadth described. Tho I will point out that the two new images that we can see (the night sky and the sunset) does have something in common - they have both been boosted/reshared by other accounts, this interaction is what probably made my server fetch those posts.
I wasn't following your pixel account before but now I do, so hopefully my server should fetch all future posts correctly....
thx. I think is best for us to use a new URL to avoid any issues. I will create in parallel another instance at a new domain and see how it goes. If it works well then we just use that and ditch the pix.trom.tf :)
Before you do that also just try posting something new, because I think once someone follows you then their server will start fetching new posts - I just followed you now on pixelfed, so let's see if my mastodon server starts picking up your new posts from here on out.....
This entry was edited (Saturday, September 26, 2026, 9:02 PM)
Yes I can see your post, see screenshot below. And this post wasn't boosted/reshared by anyone, so it shows that my instance started fetching your new pixelfed posts once someone started following your account (me in this case). So at least we know fosstodon.org is federating correctly with your instance..... I suspect that because your old instance was shut down years ago maybe that helps with the keys issue 🤔
Wait it was also an issue with our Friendica that I just fixed....so lets wait maybe it was our friendica that was unable to resolve these....that would be great.
To the tech world Rich Hickey gave a number of sublime talks on this very same theme, and called lying in the sun "Hammock Driven Development", a most worthwhile exercise. Totally agree! 🌞
Auf YouTube findest du die angesagtesten Videos und Tracks. Außerdem kannst du eigene Inhalte hochladen und mit Freunden oder gleich der ganzen Welt teilen.
Me goofing around with technology in 1996: My family: "But will this help you get a job?" Me: "I dunno, but it's fun!"
Me goofing around with technology in 2026: My family (in my head, now): "But will this help you get a job?" Me: "That's the funny thing: there are no jobs! 🤣"
The ghosts of five supernovas haunt this extraordinary image. It was acquired at Oukaïmeden Observatory in Morocco with approximately 200 hours of observations and shows a large patch of the sky, equivalent to the area of one thousand full moons tiled together, in the constellation of Auriga (the charioteer). From left to right, the five supernova remnants visible across the field are G181.1+9.5, G182.4+4.3, G179.0+2.6, G180.0−1.7 (Sh2-240, the Spaghetti Nebula), and G178.2−4.2. They are highlighted in the annotated image, together with open cluster M37 and the Tadpole Nebula. As each explosion exp
... Show more...
Woah! You think this real?
It is!
The ghosts of five supernovas haunt this extraordinary image. It was acquired at Oukaïmeden Observatory in Morocco with approximately 200 hours of observations and shows a large patch of the sky, equivalent to the area of one thousand full moons tiled together, in the constellation of Auriga (the charioteer). From left to right, the five supernova remnants visible across the field are G181.1+9.5, G182.4+4.3, G179.0+2.6, G180.0−1.7 (Sh2-240, the Spaghetti Nebula), and G178.2−4.2. They are highlighted in the annotated image, together with open cluster M37 and the Tadpole Nebula. As each explosion expanded into space, it created a growing shell of shocked gas and delicate filamentary structures shown in red (hydrogen) and blue (oxygen), respectively. These ancient stellar explosions happened independently; they are at various distances up to about several thousands of light-years away from Earth and have estimated ages up to tens of thousands of years old. Early humans may have seen them as bright new stars, fading over weeks or months.
This is from my book Rethinking Education from 2013 :) - I guessed the future quite well:
I could not foresee the many negative impacts of this "AI", but I was trying to see how these tools could help us rethink education in a saner society. The capabilities of these technologies are quite here, quite now. The incentive and hype is bad.
You know I always used to scoff at SciFi movies in the past when they showed an AI that could understand human language and respond accordingly, because nothing even close to that existed at the time. Guess I was wrong after all....
Good article about AI. I am trying to make a huge article on the AI topic for TROM. If anyone knows good sources of info, from courses to articles or researchers, please let me know. I do not want to take any sides in this, but to understand first what AI is and then how is it used and how it will impact us and the society.
Recent reports of AI systems “breaking loose” don’t suggest superintelligence. They demonstrate that AI is not intelligent enough to follow basic instructions.
I found this YouTube channel an excellent source of information on AI: youtube.com/@internetofbugs. Carl is really good at debunking the lies spread by the AI peddlers.
35 year software professional discussing software careers, quality, bugs, and AI. Special focus on LLMs generating code, and how programming might change as AI advances.
Auf YouTube findest du die angesagtesten Videos und Tracks. Außerdem kannst du eigene Inhalte hochladen und mit Freunden oder gleich der ganzen Welt teilen.
Auf YouTube findest du die angesagtesten Videos und Tracks. Außerdem kannst du eigene Inhalte hochladen und mit Freunden oder gleich der ganzen Welt teilen.
a couple tips for avoiding slop that dont require watching a groomtube vid
1. use tor browser in safest mode (knocks out A LOT).... you'll want to drap the NoScript addon back into the top toolbar though, so u can enable js on sites YOU choose, like your fediserver :) 2. do an IP address lookup before visiting any new site. If the sites IP address returns any of the C.A.G.E.M.A.F.O.A. servers just boycott it (that acronym includes the technofascists like cloudFlare, sc'amazon, groomgle, elon'gated one, microshaft, appl, falsebook, oracle and akamai. If you are clever you'll be able to download a copy of the global ip address database, its only 9 mb, and be able to check it offline. I use browser BLOKmarks to keep track of which domains i've already checked, yes 80% of websites on the internet are cancer and i wont touch them with a barge pole, saves a lot of time and energy, and you learn A LOT doing it. For example did you know that aljazeera are served from akamai (US) servers, the same US partnering woth the genocidal regime to ethnically cleanse the people neighbourin
... Show more...
a couple tips for avoiding slop that dont require watching a groomtube vid
1. use tor browser in safest mode (knocks out A LOT).... you'll want to drap the NoScript addon back into the top toolbar though, so u can enable js on sites YOU choose, like your fediserver :) 2. do an IP address lookup before visiting any new site. If the sites IP address returns any of the C.A.G.E.M.A.F.O.A. servers just boycott it (that acronym includes the technofascists like cloudFlare, sc'amazon, groomgle, elon'gated one, microshaft, appl, falsebook, oracle and akamai. If you are clever you'll be able to download a copy of the global ip address database, its only 9 mb, and be able to check it offline. I use browser BLOKmarks to keep track of which domains i've already checked, yes 80% of websites on the internet are cancer and i wont touch them with a barge pole, saves a lot of time and energy, and you learn A LOT doing it. For example did you know that aljazeera are served from akamai (US) servers, the same US partnering woth the genocidal regime to ethnically cleanse the people neighbouring the zoinist entitiy, so they can continue their land theft. Its just basic internet hygiene to check from where a site is served. M$ Goog and sc'amazon are direct partners to the genocidal state also and also one shoud never touch a site that gives them money or power.
basic hygiene that has served me well in avoiding the slop also.
Auf YouTube findest du die angesagtesten Videos und Tracks. Außerdem kannst du eigene Inhalte hochladen und mit Freunden oder gleich der ganzen Welt teilen.
Good vid! Twice quoted is: "It is difficult to get a man to understand something, when his salary depends on him not understanding it".
But it perhaps goes much deeper than that. Perhaps that technofascism has spread much further already than we are aware of. And perhaps that it has deep ideological roots, taken religious proportion. That perhaps those 'whistleblowers' are fanatical zealots of a technofascist Gestapo, where AI is a rising deity, a God, and they are Priests writing scripture in real-time. Where Heaven is to be among the elites that escape the self-inflicted destruction of our world.
AI when looking purely at the technology, has its uses, it is a power tool. But its the *perfect* tool for domination. In traditional religions people already see the hand of God in inanimate artifacts, and bow in deference. They'll easily be fooled to see a divine Being if it can formulate persuasive sentences. Become gullible to Roko's Basilisk type ideas.
In the above I make an analogy with Religion, but it is not more than analogy. Most things in the AI expert's world are neatly packed in rational scientific language. But you can wrap things in any kind of storytelling to specific targeted audiences.
Rampant surveillance capitalism combined with AI control logic can easily lead to "Bye bye democracy, hello dystopic sci-fi nightmare" with robber baron oligarchs the High Priests at the control desk.
This entry was edited (Saturday, September 19, 2026, 4:47 AM)
We had some issues with our Friendica. Unable to send the posts/comments to any remote server. Was a quick fix but took a while to figure what the issue was. Let's see if this post is sent to the fediverse...
There seem to be an issue with the Friendica instance social.trom.tf where it cannot send posts to the fediverse, it has been like that for the past two days. My friend @tio is the admin of that instance and he's investigating the issue.
Friendica 2026.05, 1595/1595, the post update version is 1550/1550. Instance https://social.trom.tf Suddenly for the past 2 days none of my posts, and perhaps from any users on our instance, can be...
The most surprising realization I had in the last years was how important it is to be public about your refusal to participate in harmful acts therefore giving others "permission" to do so as well. It really matters.
It's hard to stop other from introducing LLMs everywhere, maybe even impossible. But that doesn't mean that you can't do anything.
The first important step is to publicly say "NO", to refuse to participate. Even if you might need some software that's slopified, make your personal choices heard. Be vocal.
The best way to show people that another world is possible is by creating it. Even if just for your small community.
it screams against the insecurity of sloplickers that a lot of them take great offence at others' refusal to engage. They know about a myriad of ways in which these things are, uuuummm, extremely bad, actually. An important function of the narrative of inevitability is that it removes agency from themselves: it's an overwhelming force that happens to them, it's silly to even imagine not engaging.
But when you do, the illusion breaks and they take that as a personal, moral attack
this moral conundrum makes it difficult to talk about LLMs as an opposer
As much as the users of these systems scream that it's "just a tool" or a "purely technical decision" they know it's not. It never has been
Whatever argument you make is fighting against not logic but instead against character preservation. No matter how much you measure your words and try to deescalate to actual factual technical reasons... This is always, sill, a moral indictment. You're always "calling them out"
@VileLasagna Do I want to have nothing to do with slop because of the quality, or the fascism, or the mental damage, or the bubblenomics, or the environmental harm? All of those things. It is such a perfect storm of stay well away. But that's a lot to explain every time.
I can 100% confirm this from personal experience as well. it doesn't matter if it doesn't seem to do anything in the moment. you may learn minutes, hours, or actual *years* later that your vocal opposition made a difference for somebody.
@lrhodes Agree. I call this "saying the quiet part out loud", and the best leaders I've worked with are amazing at it. I want to get better at it myself.
I also wrote a quick post about it back in 2023, when I first noticed it as a behaviour:
“Saying the quiet part out loud” is a phrase I’ve just made up, to describe a method of building alignment on practices within a team. It’s the habit of stating why you are doing things a certain way, even when you would assume it’s obvious.
With that you create a crack in the social cohesion. Being different takes courage and voicing that even more so. You're also not alone in seeing this as "permission", I noticed the same dynamic. Shutting up never helped and one simple "I disagree" can break a dam and punch through a wall.
Don’t use bad technology, even if it’s convenient, or cheaper (for you). Take a stand, and let others know why you do that. Not to shame them, but hopefully to help them consider the impact.
I love F-Droid. I love what F-Droid stands for and I like the freedom that it gives its users. As a FOSS app maintainer I also have nothing but good things to say about the people behind the project. Maybe they’re even a bit too nice, considering how many times repro has failed due to me forgetting to commit before building 😅.
But my god is it hard to find human-written software now.
I have only one little issue with them that sometimes it's bloody difficult to find something there on their website and you keep searching using different keywords and you still end up with nothing 😐
Yeah, there are filters for f-droid to not show stuff that has even a remote link to anything commercial, or has potential to connecting to anything commercial, amongst other things. F-Droid, however, does not have any filter for AI related stuff. Which I find really sad. F-Droid does not ever require a disclosure about that stuff :(
This post hit close to home. I was always excited to take a peek at all the new apps added to f-droid but lately it's been dreading.... Every new app I see I've had to go take a close look at the repo to figure out if it's AI generated or not, and was quite disappointed to see that most of them were in fact AI slop.
Looks like a couple apps I use are ai suspect. One I pretty much knew. Some are old, and have had a development acceleration recently.
I haven't had issues with the apps I use, except the one game I knew was ai which looks sterile, and is boring.
I have mixed feelings about it. If the app is useful, and it works, why not? OTOH, it makes it harder picking the useful apps out of the flood. A lot of it is low effort crap. I was noticing before this blog.
This entry was edited (Tuesday, September 15, 2026, 11:54 AM)
Honestly: If the app works, does not trace or hurt me, AND a competent human has proofread and committed it and takes responsibility for the piece of work, I’m fine.
@david_chisnall It is rather disappointing isn't it, seems like an obvious low hanging fruit for them (to say it generously). I hope they implement such a thing soon.
I've made it a rule for myself to check the source code of every app I'm interested in before installing. If I find any contributions by Claude/Copilot/whatever, a CLAUDE.md file, openly welcoming contributions with AI or any trace of it in .gitignore, to the trash it goes.
A clean build from a committed tree proves reproducibility, not authorship. I'd check commit history and issue responses before treating code style as evidence.
dude for real... but the real human art now has more value. Also as AI seems to get sneakier [alledgedly if its not mostly all just for show, to sway the masses] i would begin to trust human made and audited software..
But really. Man made shit now has authenticity.. that makes it even better...
Can there be some way of signing a file that verifies youre human? Like a dna based signature or some shit that re compiles from your fingertip prick every time you sign? Invisi watermark
John ⁂
in reply to Kev Quirk • • •ralf tauscher
in reply to Kev Quirk • • •Kaare Mikkelsen
in reply to Kev Quirk • • •MuhammadFree🇵🇸
in reply to Kev Quirk • • •Litchi Pi
in reply to Kev Quirk • • •Thanks