C:\>DIR Global ‘Agentic Regulator’ Hackathon
Market Manipulation, Agentic Herding & Stability
Create agentic solutions that will monitor and mitigate against potential market manipulation and herding by autonomous AI agents.
Context for the Problem Space
AI-driven decision-making now dominates market activity, with algorithmic systems accounting for the vast majority of equity trading and hedge fund strategies.
Risks include but are not limited to:
Algorithmic herding and monoculture risk caused by models converging on similar strategies and data
Machine-speed amplification that turns modest shocks into rapid, large-scale market dislocations
Supply chain concentration creating single points of failure around a few foundation models and cloud providers
Detection gaps where complex AI strategies create market abuse or correlated behaviour without explicit intent
Potential Solution Areas
Mitigating Algorithmic Herding
How can regulators prevent systemic risks when independent AI models converge on identical trading strategies during market stress?
Managing Machine-Speed Volatility
How can authorities design dynamic circuit breakers to halt cascading shocks driven by ultra-fast autonomous agents?
Eliminating Data Blind Spots
How can supervisors gain real-time visibility into the specific AI models and data sources firms deploy to catch hidden market correlations?
Addressing Supply Chain Concentration
How can policymakers mitigate systemic single points of failure caused by widespread reliance on a few dominant AI vendors?
Illustrative Hackathon Prototypes
AI Adoption Tracking
A surveillance swarm that aggregates public and regulatory data to map the AI supply chain, tracking firm dependencies and shifting vendor concentration in real time.
Herding Scenario Lab
A simulation copilot that lets policy teams stress-test candidate rules against synthetic markets to quantify how specific interventions affect liquidity, correlation, and crash probability.
Correlation Sentinel
A real-time surveillance agent that monitors order-book data to detect behavioural clustering, distinguishing benign market reactions from destabilising algorithmic herding.
Agentic Collusion
A market monitoring and analysis agentic system that can detect agentic collusion conducts & behaviours to ensure market integrity and competition.
Dynamic Kill-Switch Protocol
An automated circuit breaker that monitors live trading data for acute herding, dynamically throttling or disconnecting autonomous agents to prevent systemic market meltdowns.
Reference & Resources
FSB (2025), Monitoring Adoption of AI and Related Vulnerabilities in the Financial Sector - the indicator framework and data gaps this track should target
Bank of England (2025), Financial Stability in Focus: AI in the financial system - correlated positioning, vendor concentration and herding channels
IMF (2024), Global Financial Stability Report, Ch. 3: AI and capital markets - herding, data oligopolies and market concentration
BIS (2026), Global giants in the AI supply chain
IMF (2025), Regulatory Considerations Regarding Accelerated Use of AI in Securities Markets - synchronised trading patterns and supervisory responses
Danielsson & Uthemann (2025), Artificial intelligence and financial crises - why authorities need their own AI engines, AI-to-AI links and automatic facilities
arXiv (2026), AI and Systemic Risk: Performative Prediction, Algorithmic Herding and Cognitive Dependency - unified formal model of the herding-monoculture trap
FT (2026), ‘Kill switches’ - could be needed for AI-powered trading, BoE official says.
Contact
For additional questions, please contact: contact@cdir.global