Deel

Deel Lab is a research consortium for global work and labour compliance. In collaboration with the Conflict Analytics Lab at Queen’s University, Orrick LLP, Deel Lab brings experts and data together to contribute to today's discussion around the future of global employment policy and labour compliance.

Chair, Deel Lab for Global Employment 

Conflict Analytics Lab

The Conflict Analytics Lab is a consortium for AI research on law, compliance and conflict resolution.

Director and Founder 

Cornell University

Visiting Professor

Course: 
Negotiation, Arbitration and Mediation (Labour and regulatory)
Comparative Economic Governance: EU and US Responses to the Economic Crisis

Visiting Assistant Professor of Law

Queen's Law

Associate Professor & Queen's National Scholar

Cross-appointed to Smith School of Business and Ingenuity Lab

Assistant Professor & Queen's National Scholar

Court of Justice of the European Union (General Court)

- Legal counsel at the Cabinet the French Judge 
- Rapporteur (case handler) in several high-profile antitrust and labour law cases before the Court, including one that resulted in a €500 million fine.
- Trademarks 
- Eu Civil services Cases

Referendaire

Harvard Law School

Teaching Negotiation and Conflict resolution

Affiliate Faculty Member

New York University

Research collaboration on the expansion of the legal mandate of central banks (including the Fed, the ECB, the Bank of Canada and the Bank of Israel)

Emile Noël Fellow at NYU School of Law

Conseil d'Etat

•	Researched comparative EU economic and administrative regulation (competition, pharmaceutical, financial, employment and social policy) in reference to appellate complex cases
•	Advised on Court opinions concerning government legislative proposals
•	Specialised in common law and civil law systems (US, UK, Spain, Italy and transition countries)

Legal Advisor

ESSEC Business School

Major Projects:
•    European Commission. Analyst and Trainer for ESSEC team working to assess EU social, labour policy and environmental law-making process.
•    European External Action Service (EEAS). Drafted report and case study on the mediation role of the EEAS in groundwater conflict management. Developed and designed training for senior EU diplomats.
•    OECD (Organisation for Economic Cooperation and Development). One time consultation among Mediation Experts and National Contact Points (NCPs) involved in the application of OECD guidelines for multinational companies.
•    French Ministry of Foreign Affairs. Advised and trained French diplomats on international dispute resolutions mechanisms.

Research Associate

Ecole Nationale d'Administration (ENA)

Teaching Public Dispute Resolution to Top French Civil Servants

Maître de Conferences - Négociation et Gestion de Conflits

Chair @ Deel Lab | Building Legal AI l Driving Global Work Policy

We’re hiring to expand the OpenJustice open-source community.

openjustice.ai is an open-source, no-code platform that lets lawyers fine-tune and align language models in plain language. Together with Prof. Mariam Guizani, PhD, and Xiaodan Zhu, we’re building a global community where legal practitioners can express their reasoning directly in AI systems — not just through text corpora.
Our goal is to make legal AI more dependable and introduce new ways to align models by linking lawyer-authored reasoning layers to any LLM (GPT-4, Claude, or open models). In a modest way, we’re trying to build something like a “Hugging Face for lawyers, by lawyers” — a place where the community shapes how these systems learn.
Current projects include:
⚖️ Client intake for legal aid – with Pro Bono Ontario and the Harvard Law School @A2J lab
 🏛️ Judicial AI – with courts in Canada and the EU on claim processing and deepfake evidence admissibility
 💼 Firm-knowledge alignment – with Osler, , Miller Thomson, and Deel
We’re now hiring:
🔹 PhD position – open-source ecosystems & legal AI governance:
 https://lnkd.in/eJt6-pkz
🔹 Full-Stack Developer – builder interface & integrations:
 https://lnkd.in/eUyFtaSM
If you’re interested in open, trustworthy AI and want to help shape how lawyers and courts collaborate with technology, reach out or visit https://openjustice.ai
.
#OpenSource #LegalAI #AIandLaw #AccessToJustice #Hiring

Dr. Megan Ma
Aurelia Tamo-Larrieux
Joshua WALKER
Joel Niklaus
Xiaodan Zhu
@colin raffel
Mila - Quebec Artificial Intelligence Institute
Maxime Cohen
Vector Institute
MaRS Discovery District
Renee Danser
Mandy Mobley Li

Samuel Dahan

Bringing Legal Reasoning to AI Models l Law Professor l former EU Official

I’m writing a short op-ed on the recent Quebec case where a self-represented litigant was fined $5,000 for submitting AI-generated, hallucinated case law.

If you have an opinion, please reach out or comment — I’d really appreciate hearing different perspectives
.
What struck me is less the amount of the fine than what it says about where we are.
Are we punishing the symptoms instead of the problem?

Is it just me, or is something off here?
Yes — this wasn’t a small-stakes matter (the claim was around $2.7M, with sophisticated plaintiffs).

And yes — courts are exhausted by a flood of GenAI nonsense mostly coming from lawyers.

But I still can’t shake the feeling that this sets the wrong precedent.
Globally, 5 billion people lack meaningful access to justice.
https://lnkd.in/eFm9vMxC

In Canada, around 80% of legal problems go unmet (https://lnkd.in/evh-BGwE)

In the U.S., 50% of eligible legal-aid seekers are turned away because the system simply can’t cope. https://lnkd.in/eDPvufJk

Against that backdrop, fining a non-lawyer for using the only tools they can access feels… misguided.

Courts understandably want to deter unreliable AI submissions. Lawyers absolutely should be held to high standards. But when ordinary people — who already can’t get help — are punished for turning to tools that everybody says will empower them, it sends a strange message:
“We can’t assist you, and you also can’t use anything else.”

This won’t reduce the use of AI in court. It won’t help overloaded judges. And it certainly won’t improve access to justice.

If anything, it highlights a deeper truth: people are desperate for help, and the justice system isn’t meeting them where they are.

Curious what others think — is this a necessary deterrent, or are we punishing the symptom instead of the problem?


https://lnkd.in/eCy8BRqW
CTV News
#a2j

Frontier AI for Legal Reasoning | No-Code Legal AI Design | Law Professor

Sharing the benchmarking results on how #Gemini-3 performs on Canadian #employment-law reasoning, as part of the broader #HuggingLegal benchmarkathon led by Hugging Face.

Key takeaways from the benchmarkathon:

On standard contract- and evidence-law tasks (e.g., hearsay admissibility, NDA-clause interpretation), Gemini-3 delivered almost flawless performance — well above previous models.

On more complex statutory and case-law reasoning under the Wilful Misconduct benchmark (including Canadian employment-law dismissal scenarios), Gemini-3 produced structured, plausible analyses that show real potential for practical, high-stakes domains.

However, in fact-recall tasks like AA-Omniscience, the model still answers confidently when uncertain — a reminder that expert oversight remains essential.

 benchmarkathons are an important step toward rigorous, community-driven evaluation of legal AI — work that ultimately supports safer deployment and better access to justice.

Many thanks to Hugging Face for enabling this collaboration with the OpenJustice team (led by PHD candidate Chu-Fei Luo and Queen's Law's JD Sophie Franco at the Conflict Analytics Lab.
Joel Niklaus 
David J. Doorey 

here is full article 
https://lnkd.in/eQxhDwmJ

#LauzHack will take place at EPFL next weekend, and together with my colleague Aurelia and the Code Design Lab, we’re hosting a dedicated legal stream with the CAL:

 LLM x Law — Building Legal AI for Access to Justice.
What makes this hackathon special is that it’s not only for technical teams.
 We’ve designed two tracks so anyone interested in legal innovation can take part.

1) Legal / non-technical track
 A chance for lawyers and law students to contribute directly to AI development.
 We’ll be testing the OpenJustice no-code interface, where legal professionals shape the reasoning layers that guide an LLM — encoding questions, logic, and workflows in plain language so others can refine and extend them.

We’re starting with something tangible for legal aid  :
 ➡️ Client intake 
 This builds on our work with Pro Bono Ontario in Canada.
 We hope teams will prototype a version for Caritas or other legal aid providers in Switzerland. The infrastructure is already there — it just needs legal input.

2) Technical challenges
 For CS, ML, and data-science participants, challenges include:
 • A Word/LibreOffice plug-in for legal workflows
 • A computer-vision node for document extraction (insurance, criminal filings, trademarks, etc.)
 • A programmable Swiss-law research tool aligned with internal data
 • Plus other open tasks suited to interdisciplinary teams

This mix of technical and non-technical challenges is intentional — meaningful A2J work needs both lawyers and engineers.
Looking forward to seeing what the teams build.

 If you’re at EPFL, come find us.

This is part of the broader OpenJustice open-source initiative — new models, workflows, and tools created at the hackathon will eventually be available on the OpenJustice open-source hub for others to build on.
(OpenJustice Switzerland community: feel free to reach out if you’d like to join or support one of the projects.)
Join discord channels: 
LauzHack : https://lnkd.in/ee6f_Y2B
Hugggingface Legal : https://lnkd.in/eRyMxD8x
OpenJusitce creator: https://lnkd.in/eviuTJBu

Access to justice shouldn't be a luxury, yet financial constraints and complexity often stand in the way.

We are thrilled to announce that we will be at LauzHack next weekend to tackle this exact issue! ⚖️ 💻 

We are inviting participants to "Hack for Justice" with our challenge: LLM x Law: Building Legal AI for the Public Good.

We are excited to see what the teams build using the OpenJustice platform. Big thanks to the OpenJustice team for providing the tools and on-site expertise to help the teams succeed. Samuel Dahan 

See you there, let’s hack something that matters!

It was a real privilege to partner with EPFL and the @legal design code law  community for this year’s #LauzHack.
Great to see teams using the #OpenJustice platform to prototype legal-AI tools that respond to real needs — from privacy compliance to client-facing legal processes.

Congratulations to all the participants, and to the students who built the Legal Compliance Assistant. Impressive work in just 24 hours.
Conflict Analytics Lab 
Queen's Law

At this weekend’s LauzHack, the Legal Design & Code Lab challenged teams to "hack for justice". In just 24 hours, 7 teams built innovative legal AI tools for the public good, with the goal of improving access to law and streamlining legal processes ⚖️🧑‍💻

Huge congrats to the Legal Compliance Assistant team for winning our LLM x Law challenge! 🏆🥁

🚀 What did they build? Developers often struggle to interpret complex privacy and security laws, leading to unintentional violations. This team built a VS Code Extension that performs a privacy and compliance audit on a codebase. It helps engineers build software that respects data protection and security standards while they code, translating legal requirements into practical recommendations.

Special thanks to ​​Reymond & Partners for sponsoring this prize and fueling innovation in the legal sector, and to the OpenJustice team for providing the technical platform and on-site expertise to help the teams succeed. 

LauzHack | Conflict Analytics Lab | Aurelia Tamo-Larrieux | Johannes David | Kevin Yu | Vlada Druta | Alice Palmieri | Oleg Lavrovsky | Samuel Dahan | JetBrains | FDCA UNIL - Faculté de droit, des sciences criminelles et d'administration publique | EPFL | Laia Mogas Pladevall | Biel Altimira Tarter | Pol Resina Martínez | Adrià Vico Sánchez

#LegalTech #AIforGood #DevTools #LauzHack #LegalDesign #AccessToJustice

If you have kids who spend time watching other kids play with toys, unbox gadgets, or review games on YouTube or TikTok, the EU’s new kidfluencer proposal is worth paying attention to — and not only in Europe.

I’m sharing a recent CBC piece on this topic, which I discussed briefly with the author - James Dunne, MPPA, earlier this week. The article raises questions that go well beyond Europe, and I’d like to turn this into a broader community conversation — and eventually a advocacy/policy project. If you work in this area, or simply have thoughts as a parent or caregiver, I’d really welcome your ideas and comments.


The core idea is simple: when children become online performers, they deserve the same basic protections as child actors.
But the details are where things get complicated.

Banning kidfluencing won’t work — legally or technically.

Legally, it’s hard to define who the “influencer” even is when parents run the account and children appear constantly.

Technically, detecting harmful or exploitative content is still a cat-and-mouse game. Platforms have little incentive to invest in the kind of detection systems that would be required.


Still, there are elements in the EU proposal that are worth applauding:

Stronger age and identity verification for accounts monetizing children

Independent third-party oversight

Clearer obligations for platforms to assess risks to minors

And of course, what we often call the Brussels Effect matters:
when the EU moves, global standards tend to shift.
Canada doesn’t have an equivalent framework yet, but if big platforms adjust their systems for Europe, Canadian children may benefit indirectly.

It's an important conversation, and one that will only grow as more kids (and parents) turn online visibility into work.


https://lnkd.in/edvwh8XS

The Courage to Resist AI #FOMO craziness.

Pouring billions into platform wars with almost no product differentiation — and even less evidence of sustained impact. 

It’s starting to feel uncomfortably similar to the financialized, over-heated markets we saw before 2008 — lots of capital velocity, not much underlying substance

Adding Legal Reasoning to AI Models l Law Professor l former EU Official

Founder & CEO, Curiouser.AI | Berkeley Instructor | Building Values-Based, Human-Centered AI | LinkedIn Top Voice in AI

After The Crash, Apple Will Drive By The Roadkill
And maybe even go bargain shopping

Bloomberg confirmed it two days ago: Apple’s “restraint” on AI spending mocked for two years, now looks like genius.

While the rest of Big Tech built GPU bunkers, Apple quietly built for what comes after the crash.

The Setup

Microsoft: $80B AI capex for 2025

Amazon: $100B

Meta + Google: $120B+ combined

OpenAI: Billions burned, $5B+ annual losses

Then DeepSeek proved you don’t need $100B to build frontier models.

Ready for the fire sale?

Five Reasons Apple Wins Post-Crash

1. Unit Economics

OpenAI loses money per conversation.
Apple’s on-device AI costs developers zero.

One needs infinite capital.
One is profitable at scale.

2. Supply Chain 

Apple ships 200M+ iPhones every year with 99.9% quality control across three continents.

OpenAI and Anthropic?

Never manufactured anything at scale.

When AI moves to edge devices in billions of hands, the question isn’t “Who has the biggest model?” but “Who can ship a billion units that work?”

There is only one answer.

3. Marketing Wizardry

Apple turned “megapixels” into “portrait mode.”
They turned “LTE” into “FaceTime anywhere.”

Now they’ll turn “on-device foundation models” into something grandma uses without ever knowing it’s AI.

OpenAI teaches you to prompt.
Apple ships things that just work.

4. Privacy

Google and Meta cannot copy Apple’s privacy stance without detonating their business models.

5. Platform

Apple doesn’t need the best model.
They need the best distribution.

Let OpenAI, Anthropic, Google, and Meta burn billions chasing AGI.

Apple builds the platform where useful AI quietly reaches 2 billion devices.

The Roadkill

After the crash, the landscape will be littered with:

Vaporware startups that could never reach profitability

Hardware founders who couldn’t manufacture at scale

Foundation labs with revenue that never justified burn rates

GPU cloud companies selling capacity for pennies

Apple will drive by with $162B cash, 2B devices already deployed, a supply chain refined over 30 years, and consumer trust built over 40.

Sustainable model
Operational excellence 
Consumer trust 

Apple didn’t miss the AI revolution.
They simply had the courage to resist 
********************************************************************************
“The trick with technology is to avoid spreading darkness at the speed of light.”

Curiouser.AI is backed by a growing community of believers who share a simple conviction: the next era of AI must elevate people, not diminish them.

$400,000 raised in total. 40+ investors. 60% product paid conversion.
If this resonates, visit WeFunder (link in comments).

Sources

¹ Bloomberg, Nov. 11, 2025
² The Motley Fool – DeepSeek / Apple analysis
³ Menlo Ventures – Enterprise AI Adoption Data

Should #Kidfluencers be banned (part 2)?

A quick follow-up to the discussion earlier this week on the EU’s #kidfluencer proposal — Global News also covered the issue today.

The growing interest shows how important (and complicated) this topic is for parents, policymakers, and platforms.

Thanks to Angela Kokott and shaye ganam for taking the time to look into this issue.

 https://lnkd.in/eaB_cEqA

Open article: Should 'kidfluencers' be banned? by open.spotify.com

Professor of Law @ Illinois Tech - Chicago Kent College of Law

🚨"What is an Agent? A Conceptual Primer and History of Agents and Agentic AI" -- Draft Book Chapter for forthcoming textbook on Artificial Intelligence for Law and Finance (2026 Forthcoming) ✅ 🚀

📕LINK HERE:  https://lnkd.in/gcYNemrR 

🤖 ABSTRACT: This chapter addresses the need for definitional clarity regarding "agents" and "agentic AI" amid widespread confusion in industry and academia. Drawing upon nearly a century of scholarship across eight disciplines—including philosophy, law, economics, cognitive science, and computer science—we offer a comprehensive framework for identifying and evaluating agentic systems.

Synthesizing work from Anscombe to recent advances in Agentic AI, we propose a three-level hierarchy of agency: Level 1 (Agent) is defined by the minimal properties of Goal, Perception, and Action (GPA); Level 2 (Agentic System) adds the operational properties of Iteration, Adaptation, and Termination (IAT), implemented through traditional algorithms; and Level 3 (Agentic AI) fulfills these six properties using AI or large language models (LLMs) for planning and orchestration. 

Beyond theoretical definitions, we provide a practical evaluation rubric to help practitioners distinguish genuine agentic systems from sophisticated tools or single-shot chatbots. The analysis explores critical dimensions, such as the autonomy spectrum and entity frames, to clarify the boundaries between delegated proxies and self-directed entities. Finally, the chapter outlines essential safeguards for professional deployment in high-stakes domains like law and finance, including attribution, provenance, escalation protocols, and confidentiality controls.

CC: Michael Bommarito Jillian Bommarito, CPA, CIPP/US/E

taplio

Samuel Dahan's Linkedin Analytics

Samuel Dahan

Performance ScoreSamuel Dahan's LinkedIn Performance

Samuel Dahan's Best Posts (last 30 days)

Ready to grow your LinkedIn presence?

taplio

.css-6ytdur{background-image:linear-gradient(284deg, rgba(39, 43, 72, 1) 0%, rgba(33, 47, 183, 1) 100%);color:transparent;-webkit-background-clip:text;background-clip:text;}Samuel Dahan's Linkedin Analytics

Samuel Dahan

Performance Score.css-goub6s{font-size:var(--chakra-fontSizes-md);font-weight:500;color:var(--chakra-colors-gray-600);display:block;margin-top:var(--chakra-space-1);}Samuel Dahan's LinkedIn Performance

Samuel Dahan's Best Posts .css-1emoo84{font-size:var(--chakra-fontSizes-md);font-weight:400;color:var(--chakra-colors-gray-600);}(last 30 days)

Ready to grow your LinkedIn presence?

Samuel Dahan's Linkedin Analytics

Performance ScoreSamuel Dahan's LinkedIn Performance

Samuel Dahan's Best Posts (last 30 days)