🧠 This week’s AI in Trading: Scale, Risk, and Regulation
🔹 1. Norges New AI Playbook
Norges Bank Investment Management (NBIM), is deploying AI to drive down trading costs and optimize execution. The goal: to slash $400 million annually, with nearly $100 million already saved.
Key Innovations:
- Predictive Execution Models: AI is used to forecast internal trading flows, allowing trades to be timed more strategically and minimizing market impact—especially around index-related events.
- Massive Trade Volumes: With over 46 million trades per year, small percentage gains translate into massive dollar savings.
- Cultural Shift: CEO Nicolai Tangen emphasizes that AI’s success isn’t just about tech—it’s about people. A tech-literate, innovation-first mindset is being embedded across the organization.
Read more here – https://on.ft.com/3Z27S8t
🔹 2. AI as a Short Seller’s Secret Weapon
In April, The Bear Cave newsletter spotlighted how GPT-4 helped identify 100 structurally weak companies in the Russell 2000. AI is now being used to uncover red flags—such as poor governance, misaligned executive pay, or flawed business models—with unprecedented scale and speed.
Strategic Edge:
- LLM-Driven Forensics: AI combed through earnings calls, SEC filings, and media coverage to surface patterns in leadership behaviour and underperformance.
- Augmented Equity Research: Analysts are pairing human judgment with AI screening tools to quickly filter for short candidates.
- Timing Alpha: Faster insight discovery enables more accurate entry points for bearish trades.
AI is expanding the reach of activist and fundamental short-sellers, making long-short equity strategies more agile, data-rich, and defensive in volatile markets. Read more here – https://thebearcave.substack.com/
🔹 3. Yet more from Regulators
While ESMA guidance is still pending, the recent speech by Jessica Rusu, FCA chief data, information and intelligence officer delivered at the Innovate Finance Global Summit (IFGS) 2025 introduced live testing environments for AI in trading —vital for real-world model validation. Read more here – https://www.fca.org.uk/news/speeches/ai-growth-how-fca-can-help
Policy Shifts to Watch:
- AI Sandboxes: Real-market AI testing will become a compliance necessity for algorithmic and HFT desks. The FCA plans to launch a live AI testing environment in September 2025 to enable firms to test AI tools in real-market conditions, facilitating safe and responsible AI deployment while providing the FCA with insights into AI’s impact on financial markets. Read more here – https://www.fca.org.uk/news/press-releases/fca-set-launch-live-ai-testing-service
- SM&CR Adaptation: Existing senior accountability frameworks will cover AI, reinforcing personal responsibility for model risks. Read more here – https://www.fca.org.uk/publication/corporate/ai-update.pdf
- Regulatory expectations continue to evolve. Firms need governance, explainability, and auditable logs for all AI-driven decisions—especially those related to execution and portfolio management. As per the BoE report earlier last month of “herding risk”—AI models making the same decision at the same time, especially during market stress (https://www.bankofengland.co.uk/financial-stability-in-focus/2025/april-2025)
🔹 4. Private Markets, Public Problems: AI and the $122B Secondary Boom
As secondary trading of private company shares approaches $122B in 2025, AI is becoming a double-edged sword. It powers deal discovery, pricing, and risk scoring—but also intensifies legal and reputational friction. FigureAI a robotics unicorn issued cease-and-desist letters to brokers marketing its shares without permission, spotlighting growing control issues in AI-powered secondary trading.
- Automation Meets Legal Grey Zones: AI is being used for due diligence, fraud detection, and transaction routing—often with limited regulatory guidance.
- Valuation War Games: AI-generated price signals are clashing with internal valuation narratives, especially during late-stage fundraising rounds.
Private markets are becoming quasi-public in behaviour, and AI is both the enabler and the stressor. Expect more conflicts between platforms, issuers, and regulators as AI continues to scale. Read more here – https://techcrunch.com/2025/04/29/figure-ai-sent-cease-and-desist-letters-to-secondary-markets-brokers/
5. Risk Radar: Security, Agentic Automation & Cloud Complexity
Palo Alto Networks’ Unit 42 Incident Response Report reveals emerging cyber risks in AI-powered trading environments:
- AI-Driven Phishing & Malware: Threat actors are using AI to craft targeted attacks on trading platforms.
- Insider Threats from Nation-State Actors: Especially targeting IP-rich financial firms.
- Cloud Exposure: Misconfigured AI pipelines in the cloud can jeopardize trading system integrity.
- Agentic Automation: AI agents working across trading, post-trade, and compliance functions need a “Trust Layer” to align with EU RTS 6/7 and maintain oversight.
Even though it’s an older story – for those that missed this first time around, the story from Origin resurfaced this week – AI Tool Origin Intelligence extracts critical data from bond term sheets with over 96% accuracy.
Impact Areas:
- Post-Trade Streamlining: AI automates ISIN allocation, settlement tracking, and trade matching—significantly reducing delays and operational risks.
- Pre-Trade Intelligence: AI is now parsing legal documents and investor disclosures in real time, offering traders early insights on liquidity, coupon structures, and covenants—often before this data is structured or priced in.
- Risk & Compliance Gains: For regulatory reporting (e.g., SFTR, MiFID II), AI minimizes manual intervention by tagging, classifying, and validating trades automatically.
As bond markets digitize and secondary liquidity increases, AI becomes essential for bridging fragmented data, reducing friction, and unlocking alpha—especially in complex instruments like structured credit or EM debt. Read more here – https://www.originmarkets.com/press-releases/origin-inteligence
📌 Final Takeaway
AI is not just a competitive advantage—it’s becoming part of the foundational infrastructure of trading – transforming how trades are executed, risks are managed, and capital is deployed. Yet, with this transformation comes added complexity, market fragmentation, and new systemic risks. To capture the full value of AI while maintaining resilience, trading firms must scale its use responsibly—embedding robust governance, security controls, and regulatory compliance into every layer of their AI-driven operations.
As always, thanks for reading! Let me know what you liked, didn’t and what you would like to see more of.
Many thanks
Rebecca
