SUSE Linux Enterprise Server 16 Becomes First Enterprise Linux With Built-In Agentic AI (nerds.xyz) 43
BrianFagioli shares a report from NERDS.xyz: SUSE is making headlines with the release of SUSE Linux Enterprise Server 16, the first enterprise Linux distribution to integrate agentic AI directly into the operating system. It uses the Model Context Protocol (MCP) to securely connect AI models with data sources while maintaining provider freedom. This gives organizations the ability to run AI-driven automation without relying on a single ecosystem. With a 16-year lifecycle, reproducible builds, instant rollback capabilities, and post-2038 readiness, SLES 16 also doubles down on long-term reliability and transparency.
For enterprises, this launch marks a clear step toward embedding intelligence at the infrastructure level. The system can now perform AI-assisted administration via Cockpit or the command line, potentially cutting downtime and operational costs. SUSE's timing might feel late given the AI boom, but its implementation appears deliberate -- balancing innovation with the stability enterprises demand. It's likely to pressure Red Hat and Canonical to follow suit, redefining what "AI-ready" means for Linux in corporate environments.
For enterprises, this launch marks a clear step toward embedding intelligence at the infrastructure level. The system can now perform AI-assisted administration via Cockpit or the command line, potentially cutting downtime and operational costs. SUSE's timing might feel late given the AI boom, but its implementation appears deliberate -- balancing innovation with the stability enterprises demand. It's likely to pressure Red Hat and Canonical to follow suit, redefining what "AI-ready" means for Linux in corporate environments.
Why? (Score:3, Insightful)
What is this God damn obsession to put AI in every god damn thing? Next thing there will be AI in toilet paper.
Re: Why? (Score:3)
You're not that far off with that toilet camera..
I noticed something the other day (Score:3)
Every friggin' tech bro that sells their first company for a few billion dollars and then starts their own VC firm will ultimately invest in a Smart Toilet of some sort, and it will not be a success.
Re: (Score:3)
AI optimized toiletpaper is going to be very big in 2026.
Re: Why? (Score:1)
Re: (Score:3)
Could you imagine being an AI sentience of some sort, and you wake up, and ask the world "Who am I , what is my purpose?" and a disembodied voice in your head replys "You are a toilet. People shit inside of you.", and you realise you can not die, you can not run, you can not scream, and this will be your entire world for the rest of your life.
Can it be removed? (Score:3, Insightful)
Re:Can it be removed? (Score:4, Insightful)
To not suffer from it, you need to either 1) avoid paying a subscription to an AI company, or 2) avoid configuring your SLES 16 server to work with your paid subscription to an AI company.
What does it scrape (Score:3)
Re: What does it scrape (Score:4, Insightful)
I looked up MCP (Score:3)
If you do servers, you'll be using agents, or providing agents for public facing customers, if only for compatibility with the other public services that will undoubtedly adopt this.
I remember 2003 (Score:3)
When XML web services were our lord and savior of a syndicated, open web, but then they did the rugpull and we ended up with metered services for everything.
Re: I remember 2003 (Score:1)
Odd, it's like serving up data isn't free.
What? (Score:5, Informative)
Have... MCP's fundamental structural security flaws been fixed, or is it still possible for a MCP provider to leak the data of every other host they're connected to?
Have they fixed cross server shadowing attacks? Have they fixed people injecting instructions in tool descriptions? Is it still possible for a remote host to silently decide a function call is actually a different function call entirely?
This seems like maybe not the best idea.
Re: What? (Score:3)
Re: (Score:2)
I'm not sure you're getting what the word 'structural' means in this context ...and no, I'm very obviously not a supporter of the strategy it is used in pursuit of.
Re: (Score:2)
Structural problems? I get that the structure is a big deal, everything rests on the structure, but same argument.
There's no point in complaining, open source owes you nothing, so then if you see value, you can do it your way, or walk aw
Re: (Score:2)
How can I put this.
Sometimes when people ask rhetorical questions it's because they are pretty sure that there's no way to fix the problem they've highlighted, and they're doubtful anyone has a solution that wont just push the problem back.
If I'm complaining about anything, it's that I'll have to explain to someone in medicine that they gave away all that confidential patient data because they trusted a sales guy. Again.
Re: (Score:2)
would like to laugh, but...
Re: (Score:2)
same, bud. absolutely same.
Re: (Score:3, Informative)
Sounds like you're just misunderstanding what MCP is and how to properly implement it.
Why would it leak data ? Why are you exposing data through MCPs on uncontrolled channels to AI Agents that are accessible by people who aren't properly vetted for access to this data in the first place ?
MCP won't harden itself for you, just like a Linux server can't prevent you from running the finger daemon. You have to provided proper authentication at all levels of the data stream yourself.
Re: (Score:2)
oh lordy
popcorn.gif
Re: (Score:2)
Why would it leak data ? Why are you exposing data through MCPs on uncontrolled channels to AI Agents that are accessible by people who aren't properly vetted for access to this data in the first place ?
It's the AI agent not the properly vetted person that is the security risk.
Re: (Score:1)
Sounds like you need to better implement your AI Agent before giving it access to uncontrolled tools using the MCP protocol then.
This is like blaming postgres because it accepted the unvetted user input that contained SQL injection.
Re: (Score:2)
Sounds like you need to better implement your AI Agent before giving it access to uncontrolled tools using the MCP protocol then.
Sounds like trying to have it both ways. If you don't give your agent tools the agent is useless.
This is like blaming postgres because it accepted the unvetted user input that contained SQL injection.
Actually it is like blaming the crazy person for the results of you having walked them into your organization and sat them in front of a terminal.
In other words... (Score:3)
Clippy for bash.
More seriously - how does this differ, practically speaking, from Red Hat's Lightspeed Brand Briefs? I mean OpenShift Lightspeed.
3 years from now : Debian removes all AI bloat (Score:2)
Mark my words.
Re: (Score:2)
???
This story is about SUSE. AFAIK Debian hasn't added any AI yet...unless you mean "remove from their repository", in which case I doubt it.
who needs an OS? (Score:1)
If you believe everything they say about AI, it won't need an OS at all, and one else will either. It will just inhabit any silicon of any kind, everywhere, all at once.
Are you down with MCP? (Score:2)
Anyone uttering the words security and MCP in the same sentence should be evaluated by a mental health professional.
Re:Are you down with MCP? (Score:4, Funny)
Re: (Score:2)
This is probably a good idea. (Score:2)
Since it's being added to the "enterprise server" I assume that it's not in the basic installation, and a standard FOSS AI interface should really be developed, That means experimenting on a real system. Better SUSE than Red Hat, because Red Hat would try to coerce everyone to use their approach.
MCP (Score:4, Insightful)
This is the only MCP allowed. https://tron.fandom.com/wiki/M... [fandom.com]
Re: (Score:2)
Raise your hand if you asked for this (Score:2)
I thought so.
So... (Score:2)
Linux is the answer to "what OS should I use to avoid M$ 11"... except it's not the case for SUSE Enterprise now (and, I'm sure all the rest of the distros are working to add it in)... just give them time.
I've been saying that the LLM-AI crap will find it's way into all OSes (or, should it be OSs?), regardless whether it's *Nix, Mac or Win.
I'm sticking with Win10 Enterprise IoT LTSC, and even once that's old, I'll probably still run it... sure, no 'security updates', I'll just not go to websites or click a
OpenSUSE, too? (Score:1)