It’s a win for Team Pheonix, here are the details;
Team Pheonix, from the Securities and Exchange Board of India (SEBI), has been named the overall winner of the inaugural C:\>DIR Global ‘Agentic Regulator’ Hackathon, following a global competition that attracted 942 individual builders and 336 team submissions from more than 65 countries in the Preliminary Round. With an acceptance rate of less than 11%, 36 teams progressed to the Final Build Round to develop agentic solutions for public sector, and among which, 19 teams are from or directly affiliated with regulatory authorities.
Team Pheonix took first place for its solution addressing market manipulation, agentic herding and financial stability. The team created digital twins and AI agents designed to help authorities pressure-test the potential impact of new policies on markets utilising a multi-agent system.
The winners were announced at the C:\>DIR Summit in Cambridge, UK, on 18 September 2026 by Agustín Carstens, the former General Manager of the Bank for International Settlements (BIS) and C:\>DIR Co-Convenor.
Explore all finalists and winners, including watching their presentations and demos: https://www.cdir.global/cdir-hackathon-2026-winners
The Hackathon was created to explore how agentic AI can be harnessed by regulators and policymakers to tackle six problem spaces spanning agentic advice, AI-driven fraud and scams, agentic payments and commerce, AI agents and blockchain, agentic herding and market manipulation, and know-your-agent & digital public infrastructure.
During the final build round between 1 September to 8 September 2026, the 36 finalist teams received access to sandbox environments and synthetic datasets through NayaOne and vouch.finance, alongside mentoring, knowledge sessions and technical and pitching support. Every finalist team submitted code or a working application for assessment.
Following virtual demonstrations and live voting, 260 verified regulators and policymakers representing more than 80 institutions across more than 40 countries selected the final six teams. Those finalists subsequently presented their solutions to 18 senior global regulatory leaders including the 2026 cohort of C:\>DIR Fellows, with judging criteria including the level of genuine innovation, regulatory and supervisory fit, technical plausibility, adoption readiness and guardrails.
Six global finalists recognised and shared a total prize pool of $100,000 USD:
· Team Pheonix, from SEBI in India, was awarded first place for its work on market manipulation, agentic herding and stability.
· YU-SAM POWER, from University College London, secured second place with an agentic framework designed to assess risk and identify attacks on smart contracts.
· Rwanda Anti-Scam Team, comprising participants from the National Bank of Rwanda, Bank of Namibia, Proto, FNA, GLEIF and the University of Manchester, placed third. Its solution addressed the trust gap in instant payments by using agents to support the recovery of funds for victims of fraud and scams.
· Team Axiom, from the State Bank of Pakistan, came fourth with a supervisory intelligence system for agentic payments oversight, designed to identify emerging systemic risks including agentic herding.
· Agent Reliability Profile, from the MLCommons Financial Services Working Group, placed fifth after developing a standardised, testable reliability label for AI agents, designed to help answer a fundamental question: can an AI agent be trusted?
· Probe, from the UK, took sixth place with a regulatory tool that uses AI agents to test AI-generated advice for safety and compliance.
Special awards
The vouch.finance Award was presented to Legis Labs, whose team included participants from the University of Ljubljana and the University of Cambridge. The team developed an agent passport designed to give banks the ability to check AI-agent payments before execution. The award included US$25,000 from vouch.finance.
An Honorary Award was also presented to Project Zeus, from the Superintendency of the Securities Market of the Dominican Republic (SIMV), for its approach to tackling securities and investment fraud. The team developed a human-governed, multi-agent ecosystem designed to detect investment scam advertisements, engage with suspected scammers and gather evidence to support enforcement.
Agustín Carstens, former General Manager of the Bank for International Settlements and C:\>DIR co-convenor, said: “I had the pleasure of observing each of the finalists and was very impressed by the diversity of solutions, creativity and innovation of each of the finalists. I am also confident that the depth of talent in the Hackathon extends far beyond the finalists, to the 336 teams from around the world who applied their time and talent to participate.”
Bryan Zhang, the Co-Creator of C:\>DIR and Executive Chair of Financial Innovation for Impact said: “Agentic finance demands agentic regulators. AI agents now communicate, operate and transact across platforms, sectors and jurisdictions at machine speed. We need to fathom a new paradigm for digital regulation, one where human and AI agents work closely together to supervise effectively. The C:\>DIR Global Agentic Regulator Hackathon effectively illustrated that possibility.”
The Hackathon was supported by 48 partners, bringing together regulatory, technology, academic and ecosystem organisations. Regulatory partners included the Bank for International Settlements Innovation Hub (BISIH), Global Financial Innovation Network (GFIN), Digital Regulation Cooperation Forum (DRCF), Regulator Knowledge Exchange (RKE) and Women Regulators Network. Supporters included Networks for Humanity, the Ethereum Foundation, Swift, Google, Moneybox, Ant International and Euroclear as well as GFTN, which is providing the top six winners the opportunity to attend and demo at this year’s Singapore FinTech Festival.
Be the first to comment