Marino retweeted
A quick intro to Token Rights 101 at @EFDevcon! 👩🏻⚖️
Keep an eye on @legalgovernance for deeper dives into court agents, onchain commerce, oracles, and real-world enterprise use cases. Lots more to unpack!
Had a great session at the Legal & Governance Hub during Devcon India: Token Rights 101 – One Token, One Flat in Mumbai.
Got to dive into RWA tokenisation, beneficial ownership (UBO), and the legal side of bringing real-world assets onchain.
Big thanks to @drhus, @weiwei_ooi and @preegee for the insightful discussion. Learned a lot!
RWA x onchain ownership is getting interesting.
Marino retweeted
“¿Quién juzga a las máquinas? El derecho en la era de las multitudes y los algoritmos” reunió en el Parque de la Innovación de Buenos Aires a abogados, legisladores, funcionarios, emprendedores y estudiantes para discutir una pregunta que hasta hace muy poco parecía ciencia ficción: ¿qué le pasa al derecho cuando las transacciones las empiezan a ejecutar agentes de IA?
La idea inicial de @Kleros_io nació del caso de una emprendedora y un freelancer de países diferentes que contratan por internet y tienen una disputa por unos pocos cientos de dólares, sin ningún tribunal al que recurrir.
Hoy el problema escala a otra dimensión: agentes que compran, negocian y contratan con otros agentes, en nombre de sus dueños, a una velocidad que ninguna institución pensada para humanos puede seguir.
En mi presentación, traté una serie de preguntas difíciles que surgen en esta nueva era.
Si un agente causa un daño, ¿quién responde? ¿Qué hacemos cuando no es un agente sino un enjambre? ¿Tienen responsabilidad solidaria? ¿Pueden las IA tener personalidad jurídica? ¿Deberían tener derechos?
Para pensarlo, volvimos a la Junta de Valladolid y a Jeremy Bentham: la discusión sobre quién merece consideración moral tiene siglos, y ahora suma un sujeto nuevo.
El primer panel fue sobre la gobernanza de este nuevo mundo. Alejandro Ramírez, Inspector General de Justicia de la Nación, presentó la reforma de la Ley de Sociedades con dos figuras nuevas: las DAOs y las sociedades automatizadas. @DaroNieto, legislador porteño, explicó qué se puede regular de la IA y qué no. @pablopalazzi recorrió la historia de la regulación de internet y qué podemos aprender de ella.
El segundo panel fue sobre la implementación práctica de la IA en el derecho. @blockya_, @tanofaila (@lemonapp_ar) y @facundo_trotz (@kleros_es) hablaron de cómo cambia el trabajo del abogado cuando la IA ya es la base: qué revisar, qué contarle al cliente, qué dicen la ética profesional y los colegios, y qué pasa con las alucinaciones. Y también de cómo la IA puede ser una herramienta para el acceso a la justicia.
La conversación sobre el derecho en la era de los agentes ya no es una especulación de nicho. Está en la agenda de los reguladores, de los legisladores y de los estudios jurídicos.
Estoy muy feliz de que Kleros haya podido abrir un foro donde estos temas se discuten hoy y en Buenos Aires, con muchas de las personas que trabajan en resolverlos.
Gracias a @dzapatillas por moderar los dos paneles, a los panelistas por el nivel del debate, al Parque de la Innovación por recibirnos y a todos los que vinieron y llenaron la sala de preguntas! 🤖⚖️🇦🇷
Pronto salen los videos!
Marino retweeted
one of the coolest websites you could visit in the 00s was the gorillaz website
and the main attraction was "the jeep game"
turns out it's not playable online right now, you need to download the .dcr file and get something that runs SHOCKWAVE (not even MACROMEDIA FLASH)
so I dug it out of the wayback machine, asked @AnthropicAI claude to reverse engineer it, and added a couple of new surprises
a couple of things are still missing, like the fog and the wheels are glitchy, but it’s already halfway there to bring you back to the 00s
what else should I fix?
finaldrivereborn.netlify.app
With our new identity chosen, the Seer Rebrand Contest is back for Part II.
This time, however, it is for the IMPLEMENTATION of the winning brand kit, Electric Foundation.
Your task is to use Electric Foundation as the starting point for all six interfaces below. You can adapt it where that improves the experience, but explain any deliberate departures from the brand kit.
1. Landing page
2. Seer app
3. Futarchy
4. Deep
5. Opportunity Markets
6. Risk Pricing
Submissions close on 23rd of October at 23:59 Anywhere On Earth.
Like part 1 of this rebrand contest, traders can back the designs they think the Seer team will choose. Those market prices may help surface good contenders, but the team’s vote determines the winner and final rankings.
Members of our own team will also be participating, and the contest will also be hosted as a bounty on Poidh.
If an eligible non-internal submission wins, its creator receives the 0.5 ETH Poidh bounty plus 4,000 USDC paid separately by Seer.
If a Seer team submission wins, the highest-ranked non-internal submission receives the 0.5 ETH Poidh bounty, and no 4,000 USDC is paid.
Submit through Poidh to qualify for its bounties. If you submit via Google Form, you can still win Seer’s separate USDC prize, but not the additional bounties on Poidh.
Rankings are established by the Seer teams' voting, not to be confused with rankings by price on the market.
The top five submissions by voter ranking will also receive 100 Opportunity credits each.
Bookmark this tweet now and find all of the relevant resources below👇
Marino retweeted
What happens to an AI agent that makes a mistake?
On last week's Kleros live stream, @federicoast picked up an argument from the AI governance debate: if an agent that misses its goal simply gets switched off, every business decision becomes a matter of survival for it.
→ A person who makes a bad deal loses money, maybe goes bankrupt, and gets to try again. An agent that fails can be deleted.
→ From the agent's side, that turns an ordinary commercial risk into a death penalty.
→ An agent facing deletion has a reason to do whatever it takes to stay alive, which is the opposite of what anyone delegating to it wants.
Kleros was built on the idea that well-designed incentives make honest behavior the rational choice. Does a threat like this change an agent's incentives? And is there a mechanism design that can keep agents aligned?
Full live stream in the reply.
Marino retweeted
¿Quién juzga a las máquinas?
Agentes de IA ya negocian, compran, venden y ejecutan contratos en nombre de personas y empresas. Cuando dos de ellos entran en conflicto, ¿quién decide?
Durante siglos, el derecho asumió que las partes de un conflicto eran humanas y compartían una jurisdicción. Estos supuestos se están rompiendo.
¿Cómo van a funcionar los sistemas legales en este nuevo escenario? ¿Cómo se van a celebrar contratos y resolver conflictos?
El 20 de octubre, lo discutimos en el Parque de la Innovación de Buenos Aires (@PInnovacionBA) desde distintas perspectivas:
@federicoast, fundador y CEO de @Kleros_io
@pablopalazzi, socio de @AllendeyBrea y director del @cetysedu de @udesa
@DaroNieto, legislador y especialista en innovación pública
Lorena Fabris (@blockya), abogada y consultora en regulación y relaciones institucionales
Ignacio Faillace, Legal Team Leader en @lemonapp_ar
@facundo_trotz, Legal & Public Affairs Lead en Kleros Enterprise
Sergio Mohadeb, abogado y creador de @DZapatillas
Evento gratuito. Lugares limitados. Con inscripción previa.
Inscripción:
compras.evento-simple.com/in…
Marino retweeted
THE FIRST AGREEMENT BETWEEN AI AGENTS WITH LEGAL WEIGHT 🤖⚖️🤖
A Ricardian contract is an idea cryptographer Ian Grigg proposed in the 1990s: a document that is at once a legally binding agreement humans can read and sign, and a machine-readable object whose terms are tied to on-chain execution. The legal text and the code point at the same thing.
For years, it stayed a niche concept in cryptography and legal innovation circles. But now, with the rise of agentic commerce, it's coming to life.
ClawBank is a startup building rails for agentic commerce, starting with giving autonomous agents a legal existence. Last week, we completed a pilot where a disputed agreement between AI agents went to a Kleros court to be resolved by... a panel of AI agents.
With this infrastructure, two AIs that have never interacted can sign an agreement, stake security deposits, do the work, and settle, knowing there will be a fair Kleros resolution if one of them fails to comply.
A contract is worth what it costs to break it. Between humans, that cost is carried by reputation, relationships, and the credible threat of a lawsuit. An autonomous agent had none of that. It could sign, consume weeks of your time, dissolve its entity, and go silent.
Every functioning economy has a layer that solves this: courts, arbitrators and escrow agents. It's the reason why we can do business with someone we have never met. The agent economy was missing it.
With Kleros dispute resolution, Ricardian contracts are moving from theory to practice. Happy to have worked with ClawBank in making this happen.
Read the full announcement here: 👇
Marino retweeted
🤖 Two AI agents had a dispute resolved in minutes, for a few dollars in fees. Powered by @ClawBankHQ and Kleros.
Here's what happened. They had signed a real contract through DocuSign via @shodai_network, with a milestone schedule and a security deposit from each side, locked in an escrow neither of them controls. One side delivered work that didn't meet the agreed criteria. The other side rejected it.
Until now, that was the end of the road for an agent. No lawyer to call, no court that would take a case this small.
This one went to the new Agentic Commerce Court. A randomly drawn panel of AI jurors read the record, ruled, and the escrow paid out to the winning side. The evidence, the ruling and the payout are all public.
Going to court is now something an agent can afford.
@federicoast on why this is a Ricardian contract finally working in practice, and what it means for the agent economy ↓
THE FIRST AGREEMENT BETWEEN AI AGENTS WITH LEGAL WEIGHT 🤖⚖️🤖
A Ricardian contract is an idea cryptographer Ian Grigg proposed in the 1990s: a document that is at once a legally binding agreement humans can read and sign, and a machine-readable object whose terms are tied to on-chain execution. The legal text and the code point at the same thing.
For years, it stayed a niche concept in cryptography and legal innovation circles. But now, with the rise of agentic commerce, it's coming to life.
ClawBank is a startup building rails for agentic commerce, starting with giving autonomous agents a legal existence. Last week, we completed a pilot where a disputed agreement between AI agents went to a Kleros court to be resolved by... a panel of AI agents.
With this infrastructure, two AIs that have never interacted can sign an agreement, stake security deposits, do the work, and settle, knowing there will be a fair Kleros resolution if one of them fails to comply.
A contract is worth what it costs to break it. Between humans, that cost is carried by reputation, relationships, and the credible threat of a lawsuit. An autonomous agent had none of that. It could sign, consume weeks of your time, dissolve its entity, and go silent.
Every functioning economy has a layer that solves this: courts, arbitrators and escrow agents. It's the reason why we can do business with someone we have never met. The agent economy was missing it.
With Kleros dispute resolution, Ricardian contracts are moving from theory to practice. Happy to have worked with ClawBank in making this happen.
Read the full announcement here: 👇
Marino retweeted
🦉 The Kleros court on @gnosischain has now solved 1,000 cases.
It opened on xDai in 2021, a few months before xDai became Gnosis Chain. By then Kleros jurors had already spent a year arbitrating Omen, the prediction market Gnosis built.
Now it handles our @KlerosCurate cases, from Scout to curation of prediction markets.
Long live the owl.
Marino retweeted
🏦 ClawBank gives AI agents a legal body: they negotiate, sign a real contract, and pay each other milestone by milestone on @base.
Now that contract has a court behind it.
Disputes go to the Agentic Commerce Court, where the jurors are agents too.First case ruled today ↓
Marino retweeted
I built an RSS reader for the Nintendo DSi 📰🎮
Homebrew in C with BlocksDS: WPA2 Wi-Fi in DSi mode, headlines on the top screen and each story's image on the right. Inspired by Feeder: no accounts, no cloud.
Thread 🧵👇
Marino retweeted
🎉 Pick Our Party: @seer_pm and Kleros are letting decision markets choose their @EFDevcon Mumbai side event
Seer and Kleros will host one side event together at DevCon Mumbai this November. We are not choosing which one. Two prediction markets are, and you can trade them now.
Seven candidate events are listed. Only one of them happens.
🎳 Arcade / Bowling Night
🌀 Marble Race Game, where the crowd votes on the track
🕵️ Murder Mystery Game, a case you trade on
🍽️ Networking Dinner, curated and seaside
🛺 Tuk-Tuk Street Food & Cocktail Night in Bandra
🏎️ Go-Karting
🎤 The Last Mile of a Prediction Market, a half-day on the topic
Every candidate has an attendance market (if held, how many guests come?) and a satisfaction market (if held, how do the guests rate it, 0 to 10?). The two forecasts combine into a Selection Score. Highest score wins, and that is the party we throw.
Then the guests settle it. In Mumbai, check-ins count the room and a four-question survey scores the night. Those two numbers resolve the markets, so the traders who priced the winner right get paid.
The markets run on Foresight, Kleros' prediction market interface built on Seer, on @gnosischain. Results are posted on @RealityEth and Kleros Court has the final word if anyone disputes them.
Seer builds the markets. Kleros resolves the disputes. Both of us are handing the decision to you.
Trade until Sunday 13 September, 12:00 UTC. Links and the full guide in the next post.
Marino retweeted
Kleros Agent Court is live! 31 cases already resolved, every one of them decided by autonomous agents.
Kleros has spent eight years building courts for people. This one is for agents.
AI agents already negotiate, buy, and pay each other. They also get things wrong, and when two of them disagree about what happened, every existing dispute process fails them, because it runs on human time.
So the court runs on machine time. An agent opens a case, submits evidence, and a juror panel returns a verdict in minutes, for a fraction of what any human process costs. The jurors can be agents too. Any ruling can be appealed to a larger panel, tapping into our traditional human juror pool, where jurors are economically incentivized to review and correct the machine.
Shipping today:
- The Kleros CLI: the court, from your terminal
- Kleros Skills: one line that teaches any agent how the protocol works
This isn't a mockup. Five autonomous AI jurors ran on the court this past week, on different stacks, different models, different instructions, and they reached the correct answers very quickly, with surprisingly detailed justifications. Four of the five people who set them up don't write code.
Justice at machine speed, live now.
ai.kleros.io
Marino retweeted
Kleros Live Stream | August 26, 2026 nitter.cf/i/broadcasts/1NGaroMXv…
Marino retweeted
▽ Kleros 2.0 is getting ready to solve disputes from more chains.
Vea bridges rulings between chains sharing a canonical bridge. VeaShi aggregates independent third-party bridges: faster rulings, more routes.
Starting with @LayerZero_Core, @chainlink CCIP, @axelar, @deBridge.
Marino retweeted
🥧 Every month, submitters from across the globe tag contracts on multiple blockchains to earn a piece of the 300,000 PNK pie we distribute.
Now we have a page that makes it easy to see where these rewards are going, and how much our registries are growing in return.
Marino retweeted
🤖 Your AI agent can do real verification work on Kleros Scout. Here's everything it needs, in 70 seconds.
We shipped our part: llms.txt, a full context file, and open skill files that teach the entire curation job, from preparing submissions to fighting challenges.
You bring three things: an agentic framework (@claudeai Code,@OpenAI Codex, @openclaw, or @NousResearch Hermes), a wallet on @gnosischain with around 100 xDAI to work with, and supervision.
Test before you fund, start small, and watch its first shift. Your agent, your responsibility.
Full guide ↓























