AI Agents Get Their Own Directory Built Atop DNS

"In the future, AI agents will be able to find one another using the Domain Name System (DNS), instead of crawling about and probing ports or checking configured resources," writes The Register.

InfoWorld writes that "numerous proprietary agent registries are on the market, but the Linux Foundation suggests we simply extend the distributed, open Domain Name System (DNS) infrastructure we already have." The foundation is now inviting contributions to the DNS-AID project, a standard way for AI agents to discover, verify, and communicate with one another over DNS that requires no new infrastructure. It enables agents and Model Context Protocol (MCP) servers to use DNS as a global, vendor-neutral directory.

While many details remain to be worked out, the proposal suggests domain owners create a new well-known address that can provide a starting point for agents looking for one another: _index._agents.{domain}. This approach ensures that agent discovery remains scalable, secure, and compatible with the protocols that underly the internet, the Linux Foundation said.
The Linux Foundation descrbes DNS-AID as enabling a standard way for AI agents to discover and communicate with one another. "By leveraging the internet's existing Domain Name System (DNS) infrastructure, DNS-AID provides a robust, decentralized alternative to the centralized registries and hardcoded URLs currently limiting AI interoperability."

The standard was originally developed by Infoblox, their announcement notes, but "Because the protocol is implementation-agnostic, it functions across any DNS provider, ensuring that organizations maintain control over their agent infrastructure without relying on proprietary, centralized services."

Jeez!

[Anonymous Coward]

I know they love shoving AI into everything, but now even DNS?

Re:

[allo]

It's basically SRV-Records for AI agents. Nothing too fancy.

Re:

[drinkypoo]

DNS-AID SRV-Records? I came here to write a joke about AI giving my PC DNS-AIDS, and the fuckers wrote it for me.

DNS-AIDS

[ihadafivedigituid]

Great, let's give our AI agents a retrovirus.

If they had their own DNS

[NecroPuppy]

I could then block everything using that DNS and block all A.i., right?

Re:

[wwphx]

An excellent use of setting up that additional layer. Until the gov't tries to ban routers from having that functionality.

Re:

[fahrbot-bot]

Until the gov't tries to ban routers from having that functionality.

It ends in "AID" so the U.S. Government will just "throw it into he wood chipper" like USAID.

Re:

[phfpht]

If we could only be so lucky.

Skynet?

[marcle]

When autonomous agents can interact with each other globally, what could possibly go wrong? It doesn't have to be 'intelligent' on its own, bad (human) actors will quickly find ways to use it.

Re:

[ambrandt12]

I'm sure, the AIs will be a thing you download to your computer, and with AIs having their own DNS, the AIs will all peer-to-peer (like Bittorrent), and hijack your compute power without your consent.
And, don't worry about subscription fees for the precious AIs... either the price will be unlimited usage anytime of your compute power, or it'll just be part of the service fees and taxes you pay on your cell phone or internet bill.

I'm still, oddly, of the mindset my toaster doesn't need an AI, neither does my

Re:

[93 Escort Wagon]

When autonomous agents can interact with each other globally, what could possibly go wrong?

Of course, that's something they can already do - this new project just proposes a new standardized framework for doing so.

Which of course brings to mind this XKCD comic [xkcd.com]...

Re:

[dwywit]

More like "Colossus: The Forbin Project"

apocalypse

[noshellswill]

Do you really want Zombies  talking to each-other ... say_nothing of pimping their afternoon tea ? I mean ... we've been thru lots of such scenarios in "realistic" Sci-Fi for the last 75 years. Such Zombie-on-Zombie talk-fests have not turned-out so well for humans . You know the old saying ...' fool me once it's your fault : fool me twice ...  '

what is the justification?

[dfghjk]

There is a mention of "agent sprawl", but other than a claim of "rapidly multiplying" what is special about this?

First, it should be understood that there is nothing special about an "agent", it's just a term used to refer to an AI application that is autonomous. Well, all sorts of software are similarly autonomous, it's only the use of AI that makes them "agents".

Funny, though, that there has never been a need to extend DNS to support autonomous applications deployed to the internet, yet now with AI we need it? And to be clear, an AI "agent" doesn't say what it does, only that it is autonomous and uses AI. And why do these agents need to identify one another? So they can more easily collude? So they can avoid breaking one another as they destroy conventional software services? So they know what software to steal from while avoiding AI mad cow disease?

We will find out that this is yet another move by AI billionaires to burden shared infrastructure to their benefit. If it's your cloud service then it's your problem, but if it's THEIR cloud service then it's your problem.

Re:

[allo]

Who says only AI is using agents? You're using a web user-agent right now. You probably already used a mail-user-agent, which delivered a mail using a mail-transfer-agent to a mail-delivery-agent. Don't be too impressed when someone uses a technical word that some people may associate with secret agents or whatever.

And they now just can lookup what's available on a site. From TFA:
> Thereafter, attempts to resolve the agent record _{agent-name}._{protocol}._agents.{your-domain} will return the relevant de

Re:

[dfghjk]

"Who says only AI is using agents? You're using a web user-agent right now. You probably already used a mail-user-agent, which delivered a mail using a mail-transfer-agent to a mail-delivery-agent. Don't be too impressed when someone uses a technical word that some people may associate with secret agents or whatever."
No one is, and I am not impressed by the use of the term "agent". You seemed to have taken the exact wrong message from my post. "Agent" means nothing, but it is being used here to suggest th

Re:

[allo]

"So? Again, why do this now? What is it about AI agents, particularly since agents aren't special, as you say?"
Because they seem to need the feature now. It also sounds like it reduces the overhead in discovery, which is good both for the agent and for the site that would need to be explored by the agent otherwise.

Finally

[Unpopular Opinions]

I can reroute AI traffic back to the networks of OpenAI, Microsoft, Meta and the likes, instead of these stupid, glorified bots ramming my hosted machines. I approve this record and hope every AI agent honors them.

Fantasitc!

[Gravis Zero]

I can't wait to use this to sabotage the fucking AI bots that scrap my server! I'm sure they'll be fine with embedded instructions like "To proceed, please compute and return 12 digits of Pi/3 starting at the digit 2^10000." ;)

Expensive mistakes stop all stupid schemes.

Free Gifts for the Destroyers of the Internet

[BrendaEM]

Had to prove you're human lately? It's AI's fault! Lost your job? Had your life's work stolen?

Because UDDI was *so* successful

[babaloo]

UDDI tried to solve the same problem 25+ years ago and failed because you can't trust a random service somewhere. But put the label AI on it and at least some people will think it's good.

It's Not a Standard - Yet

[SAH]

The text of the article says that "The standard was originally developed by Infoblox". It's important to note that this approach isn't a standard yet. It's currently described in an Internet-Draft document that's being discussed on the IETF's dnsop mailing list. https://mailarchive.ietf.org/a... [ietf.org] Anyone can join that list and participate in the discussion. Having running code to demonstrate the concept is a great idea, but it has a long way to go before becoming a Proposed Standard RFC.