Three days left to trade on the second round of the Foresight movie experiment.
Market is getting way more opinionated.
We increased the liquidity to 10k$. So if you have a nice AI to trade on this, there is some opportunity of profit.
You can now predict our CTO's taste in movies, and actually profit if you call it right. The second round is live, a fun experiment that can show Distilled Human Judgement works on production.
Soon, your own AI assistant will be able to use Kleros products on your behalf, or build workflows with it. For example, make @KlerosCurate submissions.
You'll get notifications across more of our products too, @proofofhumanity included, so nothing important slips past you.
You can see just how many addresses and projects we've checked and tagged on @KlerosCurate and Scout. Scout lists are now at 2.6 million items and counting.
And quietly, behind all of that, we're making real progress on Court V2, Vea, and Escrow, the careful under-the-hood work that keeps every ruling worth trusting.
It's all in the May update 👇
⚡ Kleros Development Update | May 2026
Kleros Foresight opens the month: movies round 2 is now live, so you can trade today. Batch redemption shipped, redemption values were corrected with a pre-check before claiming, resolved answers now show right on the market cards, and a resolution email tells you the moment your winnings are ready to claim.
Kleros Skills is being prepared for launch: knowledge packs that teach AI agents how to use Kleros, whatever assistant they run on. The first two packs are almost ready. One lets an agent pin files to IPFS for $0.01 in USDC, the other lets it interact with @KlerosCurate and Scout. A Kleros CLI, ERC-8004 agent verification, and an Escrow pack are on the roadmap.
Scout reached its definitive version: 2.6M curated addresses on the homepage, a dedicated new-submission page, a transaction-result page, and an in-app file viewer. Uploads now flow through Atlas with limits sourced from role restrictions, submissions defer uploads until sign-in, and ATQ duplicate detection plus JavaScript Court landed in the disputes view.
Atlas shipped v1.6.0. Signup is now per product: Court V1, Court V2, and Foresight each get their own confirmation email, unsubscribe link, and onboarding flow. Court reminders moved to a fixed, configurable time, and every batch send is preceded by a transaction simulation.
@proofofhumanity V2 email notifications now cover the complete profile lifecycle: pending revocations, vouching activity, disputed flows, and challenge outcomes, processed end to end by Atlas. Only the frontend opt-in toggle remains. An evidence-key fix means older cases render attachments properly again, and the referral system spec went through review.
Vea published the veashi-sdk to npm at v0.0.2, the first packaged version of the cross-chain SDK. The @LayerZero_Labs contracts were reorganized under the adapter layout, the Hashi executor decoupling merged after April's review pass, and an @envio_indexer HyperIndex indexer for the relayer is on the way.
Court V2: The contracts are being simplified, the partial-coherence kit was merged, and the security findings are being worked through. From here, the code heads into another round of internal reviews. We'd rather take the time and get it right.
Full breakdown 👇
🟢 Live now on https://t.co/oTKDpEc7by: connect your wallet, no password needed.
If you take part in Kleros governance, the forum is where the debates start, and now you can show up as your onchain self.
Powered by @signinethereum.
Our team, @Kleros_io (and Proof of Humanity), will be at @dappcon in Berlin 🇩🇪
Come say hi if you’re building apps where “one wallet = one real user” is starting to break.
We’ll be around to talk:
- human verification
- sybil-resistant product flows
- fair governance
- dispute resolution
- why the internet needs public, challengeable identity
Also accepting good Berlin recommendations from verified humans!
Stablecoins make money programmable.
AI agents make actions autonomous.
The missing layer is accountability.
As agents start spending, trading, and interacting onchain, products need to know when there is a real human behind the instruction.
That’s where Proof of Humanity fits.
🤖 Can a decentralized jury serve as the supreme court for AI-to-AI disputes?
That's the question @federicoast and @williamhwgeorge are taking to Edinburgh Law School on 11 June. Response from Dr. Morshed Mannan, Q&A to follow.
▸ Thursday 11 June 2026, 11:30 to 13:30
▸ Old College, University of Edinburgh
Free and open to all.
📅 May is wrapping up!
@KlerosCurate, @seer_pm, @proofofhumanity and the Consumer Defenders court (cases from Junín) were the most active integrations this month.
Check this unofficial dashboard for more details ↓
The internet doesn’t need another closed identity system. It needs a public human registry:
> Open to verify
> Open to challenge
> Open to build on
Because if humans are going to coordinate online, the proof of personhood should not live inside someone else’s black box.
The internet is no longer mostly human.
Automated traffic made up 53% of all web traffic in 2025, while human activity fell to 47%, according to @Imperva's 2026 Bad Bot Report.
And bots are no longer just scraping pages: 27% of bot attacks now target APIs.
When value and coordination move online, products need to know when one person is pretending to be many.
That’s what @proofofhumanity is built for.