Linux memory management expert Lorenzo Stoakes at Arm who has been contributing to the kernel since 2014 is stepping away from some reviewership roles within the Linux kernel due to all of the noise from the flood of kernel patches resulting fro AI/LLMs.
Lorenzo announced today he is stepping away as a reviewer for the memory management (MM), core, misc, and Rust areas due to all the noise caused by AI. Lorenzo will continue serving as maintainer for areas of the kernel he oversees but is looking to lighten his load due to the increased patch burden in 2026.
Lorenzo commented in the patch series removing himself as reviewer for those areas of the kernel:
"Given the current throughput in the kernel resulting from AI-generated series, I unfortunately lack the bandwidth to generally be able to offer review except in those areas I maintain.
I have separated the patches into areas maintained by David, Andrew, and Alice + Danilo, so each can take the patches relevant to the subsystems, for convenience.
Apologies for doing this, but given the current trends I'd like to do something to make my inbox a little more manageable :)"
Lorenzo has been a longtime contributor to the Linux memory management code and also author of the book The Linux Memory Manager.
AI found a bunch of security issues that up until now had not been found. Just because no one had found them doesn’t mean the security exposure wasn’t there. Talk about shooting the messenger (AI in this case). Lorenzo should ask himself why the vulnerabilities hadn’t been found until now.
i don’t know how you nterpreted that article, but to me it was just a dude overwhelmed by volume, and sending some to others so he can continue to be effective. i don’t think he was claiming anything about AI or the patches themselves.
You may very well be right and I was being influenced by the general hatred of AI in Lemmy. As someone who volunteers my time, I can relate when the commitment changes exponentially while being completely out of my control. Thanks for pointing that out.
Are you serious? He’s one human being. He’s not getting paid to do this. All the people who volunteer their time to maintain Linux have lives and jobs too. He didn’t “shoot the messenger” He said "Given the current throughput in the kernel resulting from AI-generated series, I unfortunately lack the bandwidth to generally be able to offer review except in those areas I maintain.
I have separated the patches into areas maintained by David, Andrew, and Alice + Danilo, so each can take the patches relevant to the subsystems, for convenience.
Apologies for doing this, but given the current trends I’d like to do something to make my inbox a little more manageable :)"
It’s right there in the post. It doesn’t matter if the vulnerabilities are valid or not. The man can’t keep up with doing everything he’d been doing because its just too much work for one person to handle.
There seems to be a lot of entitlement in the open source consumer community. It’s kinda wild how so many people seem to think they can dictate how people spend their personal time just because they (wannabe dictator) like the thing they (the person spending their free time making free shit) made.
I think it’s because of capitalism as cliche as that sounds. People are so used to the idea that they’re entitled to things and stuff like “the customer is always right” that they’ve got it in their head that the consumer is sacred. That they have rights or something as the user. That people who make things should be grateful they use them. Rather than the other way around being the case. If you use FOSS software and you never donate to any of them, never submit issues, never provide diagnostic data, etc then you are just a leech. You aren’t entitled to anything. There’s nothing wrong with leeching off something like this that’s why people make it but for fucks sake be grateful. Even a little bit of money makes a difference btw. I use Debian and I will send them 5-10$ a year. I’m also considering sending them some of my old hardware but I’m not sure if it would be very useful to them lol.
Yeah, I think you’re right. I think it’s also created a bunch of entitled paying customers, too, eg the ones that think screaming at front line employees is going to do dick all to help them. I’m happy to see more businesses in my area with signs saying they’ll kick out abusive customers and hope it becomes the norm.
The hilarious (because it’s so outrageous) part is they trained several generations that the customer is supposed to be the boss but then leave hired help to actually deal with those customers, including enforcing policies that go directly against customer wants. And then if someone manages to find their contact info to complain directly to them, they usually bend immediately because they don’t want to deal with them either. But they only bend for that specific case while all the other similar ones still get the previous policy.
I agree broadly, but how do we know he isn’t being paid by ARM for this work?
Even if he is he was being paid for the workload he was doing before. That workload has vastly increased. It would not be reasonable to expect him to take on a bunch of new work with no pay increase. And that’s assuming its even possible for him to do it all by himself.
yes i can also point an LLM on the kernel, instruct it to find vulnerabilities and it will invent one because it’s a sycophantic removed. Doesn’t change the fact that 80% of those reports are absolutely worthless and that are also written by LLM are miserable to read and verify.
This is actually a great opportunity for you to bolster your argument on multiple fronts.
Grab your favourite LLM , point it at that article and ask it to summarise the meaning.
See how that stacks up against the take you just posted.
For bonus points, also provide the prompt, name of model/harness used and the response you were given, for independent verification.
That way, you get to promote your favourite workflow and provide corroborating evidence of interpretation.
your last sentence is not like the others