WorldCoin's newest pitch: Scan your eyeballs to prove AI agents really represent you
- Reference: 1773779168
- News link: https://www.theregister.co.uk/2026/03/17/worldcoin_identity_ai_agents_whatever/
- Source link:
World, known as [1]WorldCoin until late 2024 when AI became trendier than cryptocurrency, [2]announced on Tuesday that it was opening a limited beta of its new AgentKit. The new tech, says World, will serve as a way to tie AI agents directly to a human to prevent bad actors from abusing agentic AI and "infuse trust into the system."
Given this is a World venture, that damn orb is still involved.
[3]
For those that don't recall, WorldCoin spun up in 2019 with the goal of bringing everyone into its global cryptocurrency ecosystem using an [4]electronic orb that scans a user's iris and creates a unique World ID associated with that individual. The idea behind the whole thing was to put personhood on the blockchain in a pseudonymous manner that would allow individuals to transact with other World users without having to reveal their identity.
[5]
[6]
As we noted in 2023, Altman and World had a professed goal of getting a billion people on the platform within 2 years. It's been more than that, and according to the AgentKit announcement, World has only managed to sign up 18 million suckers users to date.
This is my bot. There are many like it, but this one is mine.
ONow, with cryptocurrency having largely fallen out of favor (including WorldCoin - the coin itself has [7]lost 76 percent of its value since launching in 2023) and a bunch of biometric data on file, World is looking for something to do with it. Enter AgentKit.
AgentKit serves as an extension to Coinbase's [8]x402 protocol , which allows cryptocurrency users to exchange digital cash directly over HTTP.
x402 has also been extended to serve as a way to limit AI agent access to online resources by charging them micropayments. While that's enough to filter out some bad actors, World argues that actual identity verification is needed.
[9]
"Through World ID, a person can cryptographically and anonymously prove that they are a unique human without revealing any personal information to anyone," World said in a press release. "That same proof can now extend to their agents."
[10]Rogue AI agents can work together to hack systems and steal secrets
[11]Altman: You think AI is wasted energy? Try raising 100 billion humans
[12]AI agents abound, unbound by rules or safety disclosures
[13]Google unveils master plan for letting AI shop on your behalf
According to World, AgentKit allows verified World ID holders (i.e., those who've had their irises scanned by an orb) to delegate their World IDs to AI agents, essentially serving as cryptographic proof of the individual behind the agent. A single human is allowed to delegate their World ID to as many agents as they want.
World describes AgentKit as being useful for spam prevention and other forms of abuse, like using AI agents to scalp tickets or reservations, prevent the flooding of news rankings with garbage, and the like. Presumably, they'll need to get some businesses on board requiring AgnentKit identity for certain services before consumers will take the bait, thereby finally making World relevant.
AgentKit details are [14]available on World's documentation site for those who wish to wade into the weeds and learn more.
Neither World, nor its parent company Tools For Humanity, responded to questions for this story. ®
Get our [15]Tech Resources
[1] https://www.theregister.com/2023/05/16/worldcoin_fundraising/
[2] https://world.org/blog/announcements/now-available-agentkit-proof-of-human-for-the-agentic-web
[3] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_security/front&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2abndFfrxIZ12MyHd1oz-TwAAAxM&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[4] https://www.theregister.com/2025/05/04/sam_altman_startup_world/
[5] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_security/front&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44abndFfrxIZ12MyHd1oz-TwAAAxM&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[6] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_security/front&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33abndFfrxIZ12MyHd1oz-TwAAAxM&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[7] https://coinmarketcap.com/currencies/worldcoin-org/
[8] https://docs.cdp.coinbase.com/x402/welcome
[9] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_security/front&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44abndFfrxIZ12MyHd1oz-TwAAAxM&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[10] https://www.theregister.com/2026/03/12/rogue_ai_agents_worked_together/
[11] https://www.theregister.com/2026/02/23/sam_altman_ai_efficiency/
[12] https://www.theregister.com/2026/02/20/ai_agents_abound_unbound_by/
[13] https://www.theregister.com/2025/09/16/google_unveils_masterplan_for_letting/
[14] https://docs.world.org/agents/agent-kit/integrate
[15] https://whitepapers.theregister.com/
18 million?
Were they offering free fentanyl with every scan?
Eyeball scanning doesn't work.
Well-known idea, but eyeball scanning is a bad idea. Eyeballs aren't that stable, and if you loosen recognition enough to allow variability, you lose security.
Re: Eyeball scanning doesn't work.
I thought iris scanning (I assume that's what this is) was similar to Apple's Face ID (or other depth mapping facial ID products) where it is around 1 false match per 50,000 people, or at least in that ballpark. So long as it is only being used to prove you're you that's fine - assuming you have some method of preventing brute force attacks by hundreds of thousands of people. Which is the case for Face ID because no one is stealing phones and then holding them up to 50K people to find that one false match.
I'm not sure how this works with that orb thing. If it linked to your specific orb then great that's secure (unless you are protecting tens of millions of dollars in which case stealing your orb and holding it in front of a stadium's worth of people to find a false match is worth the trouble) but what happens if your orb is lost or breaks? There needs to be a way to get you back in, though that could be done with some sort of long enough to be highly secure recovery password stored in a safe.
I suspect this is all academic as it is IMHO likely that there are some basic security screwups in this product, because it has always been vaporware so no one has been trying to attack the way they've been trying to attack iPhones for nearly 20 years.
Scammers gonna scam ...
e.g. "Let's just string together all of my failed/failing investments, where no one really understood how they were supposed to make money, and mix them into my AI investment , where also nobody - including me - understands how this is supposed to make money.
Then, give it a catchy name, run up a short hype via social media and have new "investors" buy this franken-investment until everyone understands its toxicity.
Same concept as the "Macrohard" crap of that other AI Bro last week...
No.
Just no.
"World"
Could they have come up with a more megalomaniacal name?
Anything in a Concise Dictionary, never mind a "My First Dictionary" should be inadmissable as a company / brand name.
A secure token
Until it gets stolen, uploaded or otherwise misused and cannot be replaced.
Maybe we should first invent Minority Report's eye swapping surgery and never tell the bad guys you can do that.
Re: A secure token
By next year AI images generated from billions of 50+megabit selfies will be good enough to fool any bio recognition software
Close call
- We need to scan users' balls.
- Why?
- What do you mean why?
- It's gross.
- Are you saying your balls are gross? Show me.
- I am a woman.
- So? Everyone has balls these days.
- I don't.
- Right.
*pause*
- How about eyeballs?
...