@bekkaboo@girlcock.club Linux itself is heavily using AI models. Therefore, every Linux distribution moving to recent kernel versions is heavily built on top of using AI models. Creating a Linux distribution not heavily using it would require a hard fork of the Linux kernel and many other projects.
It's unclear what would be accomplished by banning AI for a tiny portion of the code while continuing to use Linux, AOSP, Chromium and hundreds of other projects heavily using it. We'd still be benefiting from it.
GrapheneOS is currently defending its use of AI coding tools on Mastodon against complaints by various accounts claiming to be users.
We do not understand where you’re coming from or why you’re so incredibly angry with us. It’s not justified and does not make sense.
Ssshh…Most of those having a knee-jerk “AI BAD! YOU BAD!” reaction don’t have any clue WTF an IDE is, how code completion works, nor the fact that by most metrics 95%+ of code now has “AI” in the chain…and has had that for YEARS.
Here’s a crazy idea…how about instead of “AI BAD! ME HATE AI” how about some nuance? Assume ALL software in 2026 has had AI assistance, and review it on its merits.
As for the devs at GrapheneOS directly…why were you (the general you) trusting them before? Humans can be plenty dicey in ways that affect code all on their own.
Supposedly wonderful “human artisanal code” has plenty of fuckery.
These are supply chain attacks and in the case of xz utils, the attacker had gone to extreme lenghts to hide the attack from a well-meaning, good-hearthed but overworked and burnt out solo maintainer.
To compare this to bugs that people unwittingly introduce in normal human-written code is not sincere.
That’s the thing about “pure human slop”: it doesn’t need to be malicious to be catastrophic.
The second link is particularly salient - kills the “supply chain attacks are special” argument because it is precisely about “unwitting bugs in normal human-written code just happen”
Okay, but they still happen with orders of magnitude less frequency than bugs in AI code. Consider that the time between when rsync first adopted LLM-generated code and users en masse reporting rsync internal protocol errors during a backup was on the order of months.
I don’t recall that one…but in fairness…AI generates a metric shit ton more code than humans. We’d have to normalize the results.
Interestingly, looking it up now, someone DID normalize for that very case. Bug rate per commit for the AI-assisted versions landed within normal historical range. A pre-AI release had more regressions. The 3.4.3 regressions were primarily from the CVE security patches, not the AI work. Zero CVEs from the Claude-assisted commits.
The author of that article admits that the sample size is not large enough to draw meaningful conclusions.
But besides that, I believe LLM code generators can be a useful tool, provided you are willing to go over their output with a fine-tooth comb and assume it is broken until you have proven otherwise, because the hallucination problem is inherent to the technology and they’re never going to completely solve it, and are willing to overlook the myriad ethical issues with all major LLMs in existence today.
The author of that article admits that the sample size is not large enough to draw meaningful conclusions.
Hey, you brought it up - I just pulled the thread. Not my fault it cuts against your argument.
But besides that, I believe LLM code generators can be a useful tool, provided you are willing to go over their output with a fine-tooth comb and assume it is broken until you have proven otherwise,
So, exactly like a junior dev?
going to completely solve it, and are willing to overlook the myriad ethical issues with all major LLMs in existence today.
That’s a different claim than the one you opened with though. Most of those objections have documented counter-arguments, btw:
Data centres were polluting long before LLMs arrived. Crypto, cloud storage, Netflix, YouTube, AWS - AI isn’t all data centre load. Blaming the latter for the former is like blaming sunscreen for melanoma.
I’d really like to see a source for the first guy’s numbers, especially how he accounts for things like training and water usage at the power plant, and as for the second guy, Anthropic is not preserving shit. The Internet Archive preserves books. Anthropic scans them and doesn’t publish the scans so that they can train an LLM that might or might not be able to regurgitate some fragments of that text, and publish that. That’s preservation in the same way that painting a picture of you is keeping you alive forever.
Also, neither of those address the effects on creatives’ livelihoods or the mental health of LLM users. Chatbot psychosis is real. People who routinely use LLMs to do things provably become worse at doing them without them. Students use LLMs to make an end run around having to learn everything they need to know to be effective members of a society, like how to articulate their points, how not to fall for rhetorical traps, what history was really like, and between that and the disastrous effects of No Child Left Behind, teachers are quitting in droves and there’s a literacy crisis.
really like to see a source for the first guy’s numbers
The citations are inline hyperlinks throughout the piece - IEA, Lawrence Berkeley National Lab, MIT Technology Review, Google’s own efficiency data.
They’re there. Click them.
Also, neither of those address the effects on creatives’ livelihood
We’ve now gone from “AI can’t code,” to “AI code is a malicious risk,” to “humans would never do that,” to “that’s a disengenous example” to “ok, but what about this, this and this”
We’re verging on a gish gallop at this point, so I demur. Let’s stick to the claims at hand instead of litigating shifting goal posts.
On the topic of the second article -
What the judge ruled was that it qualifies as fair use specifically because they destroyed the copies - the destruction is what made the scanning fair use, not what made it wrongful.
Retaining the digital files without destroying the physical copies would have been the violation.
Meaning the law perversely incentivised this behavior. (If you want the actual court reporting, the Ars Technica piece the article quotes is the cleaner source).
Ssshh…Most of those having a knee-jerk “AI BAD! YOU BAD!” reaction don’t have any clue WTF an IDE is, how code completion works, nor the fact that by most metrics 95%+ of code now has “AI” in the chain…and has had that for YEARS.
https://github.blog/news-insights/research/survey-ai-wave-grows/
https://survey.stackoverflow.co/2025/ai
OMGWTFBBQ!
Here’s a crazy idea…how about instead of “AI BAD! ME HATE AI” how about some nuance? Assume ALL software in 2026 has had AI assistance, and review it on its merits.
As for the devs at GrapheneOS directly…why were you (the general you) trusting them before? Humans can be plenty dicey in ways that affect code all on their own.
https://www.reddit.com/r/PrivacyGuides/comments/13s7mv3/why_i_deleted_grapheneos_louis_rossmann/
https://factually.co/fact-checks/technology/louis-grossman-grapheneos-drama-c1f2ae
Supposedly wonderful “human artisanal code” has plenty of fuckery.
https://en.wikipedia.org/wiki/XZ_Utils_backdoor
https://thehackernews.com/2024/03/urgent-secret-backdoor-found-in-xz.html
https://simonwillison.net/2026/Mar/31/supply-chain-attack-on-axios/
https://www.hivepro.com/threat-advisory/axios-npm-supply-chain-attack-what-you-need-to-know/
These are supply chain attacks and in the case of xz utils, the attacker had gone to extreme lenghts to hide the attack from a well-meaning, good-hearthed but overworked and burnt out solo maintainer.
To compare this to bugs that people unwittingly introduce in normal human-written code is not sincere.
Very well. Here -
https://www.debian.org/security/2008/dsa-1571
https://www.finnie.org/2024/05/13/i-discovered-the-debian-openssl-bug/
That’s the thing about “pure human slop”: it doesn’t need to be malicious to be catastrophic.
The second link is particularly salient - kills the “supply chain attacks are special” argument because it is precisely about “unwitting bugs in normal human-written code just happen”
Okay, but they still happen with orders of magnitude less frequency than bugs in AI code. Consider that the time between when rsync first adopted LLM-generated code and users en masse reporting rsync internal protocol errors during a backup was on the order of months.
I froze my rsync package the day that was announced. fuck the dev :(
I don’t recall that one…but in fairness…AI generates a metric shit ton more code than humans. We’d have to normalize the results. Interestingly, looking it up now, someone DID normalize for that very case. Bug rate per commit for the AI-assisted versions landed within normal historical range. A pre-AI release had more regressions. The 3.4.3 regressions were primarily from the CVE security patches, not the AI work. Zero CVEs from the Claude-assisted commits.
EDIT: Correct URL https://alexispurslane.github.io/rsync-analysis/
The author of that article admits that the sample size is not large enough to draw meaningful conclusions.
But besides that, I believe LLM code generators can be a useful tool, provided you are willing to go over their output with a fine-tooth comb and assume it is broken until you have proven otherwise, because the hallucination problem is inherent to the technology and they’re never going to completely solve it, and are willing to overlook the myriad ethical issues with all major LLMs in existence today.
Hey, you brought it up - I just pulled the thread. Not my fault it cuts against your argument.
So, exactly like a junior dev?
That’s a different claim than the one you opened with though. Most of those objections have documented counter-arguments, btw:
https://blog.andymasley.com/p/a-cheat-sheet-for-conversations-about
https://aicentral.substack.com/p/why-anthropic-burned-the-books
Data centres were polluting long before LLMs arrived. Crypto, cloud storage, Netflix, YouTube, AWS - AI isn’t all data centre load. Blaming the latter for the former is like blaming sunscreen for melanoma.
I’d really like to see a source for the first guy’s numbers, especially how he accounts for things like training and water usage at the power plant, and as for the second guy, Anthropic is not preserving shit. The Internet Archive preserves books. Anthropic scans them and doesn’t publish the scans so that they can train an LLM that might or might not be able to regurgitate some fragments of that text, and publish that. That’s preservation in the same way that painting a picture of you is keeping you alive forever.
Also, neither of those address the effects on creatives’ livelihoods or the mental health of LLM users. Chatbot psychosis is real. People who routinely use LLMs to do things provably become worse at doing them without them. Students use LLMs to make an end run around having to learn everything they need to know to be effective members of a society, like how to articulate their points, how not to fall for rhetorical traps, what history was really like, and between that and the disastrous effects of No Child Left Behind, teachers are quitting in droves and there’s a literacy crisis.
The citations are inline hyperlinks throughout the piece - IEA, Lawrence Berkeley National Lab, MIT Technology Review, Google’s own efficiency data.
They’re there. Click them.
We’ve now gone from “AI can’t code,” to “AI code is a malicious risk,” to “humans would never do that,” to “that’s a disengenous example” to “ok, but what about this, this and this”
We’re verging on a gish gallop at this point, so I demur. Let’s stick to the claims at hand instead of litigating shifting goal posts.
On the topic of the second article -
What the judge ruled was that it qualifies as fair use specifically because they destroyed the copies - the destruction is what made the scanning fair use, not what made it wrongful.
Retaining the digital files without destroying the physical copies would have been the violation.
Meaning the law perversely incentivised this behavior. (If you want the actual court reporting, the Ars Technica piece the article quotes is the cleaner source).
https://arstechnica.com/ai/2025/06/anthropic-destroyed-millions-of-print-books-to-build-its-ai-models/
So, it’s more complicated than “Anthropic are book burners.”