All Things AI in Trading – 14th July 2025 – Here’s what I’ve learnt this week
From AI transforming research to agents driving trades, the FCA sharpening Sandbox initiatives, the EU tightening rules for models like ChatGPT and Gemini—and all eyes now on the potential for global regulation – here’s what I’ve learnt this week on AI in Trading
1. Markets in Motion on the Back of AI
Markets remain in flux as AI continues to reshape the investment landscape. Investors poured $10.2 billion into global equity funds in the week ending July 9, fueled largely by enthusiasm for AI and technology stocks (Reuters story). Meanwhile, asset managers like AllianceBernstein, JPMorgan Asset Management, and BlackRock are rapidly embedding AI into their investment processes to automate research, reduce bias, and generate real-time insights—with about 75% of analysts at AllianceBernstein now relying on AI tools (Business Insider story). Yet caution persists: Gappy Paleologo of Balyasny Asset Management warns that while tools like ChatGPT excel at coding and data summarization, they still lack the nuanced judgment essential for effective stock picking. He predicts AI will transform how investors interact with data and may eventually replace traditional research terminals, but insists that genuine investment insight remains fundamentally human (Business Insider story).
2. AI Moves from Research to Execution
AI is shifting from supporting research to directly driving trading strategies. Man Group’s quant equity arm has launched AlphaGPT, an agentic AI system capable of autonomously generating, coding, and backtesting trading ideas. Developed by Man Numeric in Boston, AlphaGPT scans vast datasets for new signals, writes trading strategy code, and tests ideas against historical data. While human oversight still checks outputs for errors like hallucinations, several dozen AI-generated signals have already been approved by Man Group’s investment committee for live trading—a major step toward AI executing end-to-end trading processes in the quant world (Bloomberg story).
3. FCA’s AI Sandbox – Testing Update
The FCA’s 9 July 2025 update offered new specifics on its AI Lab initiatives. Firms will be able to begin testing early-stage AI proofs of concept in the Supercharged Sandbox from October 2025, with clear application criteria outlined for participation (FCA Supercharged Sandbox details). Applications for AI Live Testing are open until 20 August 2025, requiring firms to show active AI use and readiness for collaboration with the regulator (AI Live Testing Terms of Reference). The FCA also confirmed that the AI Spotlight remains open for showcasing real-world use cases, and it is reviewing feedback gathered through the AI Input Zone, which closed in January, to guide future regulatory policy. Additionally, the FCA highlighted recent research on AI applications, bias, and findings from its AI Sprint event (FCA AI Lab publications). These updates signal the FCA’s transition from planning to execution, offering firms concrete pathways to innovate under regulatory supervision.
4. Europe Tightens AI Rules for Chat GPT & Gemini
The EU has released its finalized voluntary Code of Practice under the AI Act, targeting general-purpose AI models like GPT-4 and Google’s Gemini. The code sets requirements for transparency, copyright protection, independent risk assessments, and technical measures to prevent AI models from replicating copyrighted content. Industry groups warn the rules could hamper competitiveness, but EU officials insist they’re essential for safe and transparent AI deployment. Meanwhile, debates continue over whether stricter regulations for high-risk AI applications, such as facial recognition, should proceed as planned in 2026 or be delayed to accommodate rapid technological advances (FT story).
5. A Global Patchwork — But Momentum Grows
Globally, AI regulation remains fragmented. In the US, over 45 states have passed more than 550 AI-related bills covering privacy, safety, and bias (Financial Times, NCSL), yet there’s still no overarching federal framework, leaving agencies like FINRA, the SEC, and the GAO to integrate AI oversight into existing compliance regimes (Orrick). Australia is progressing with voluntary AI safety standards (State Street, Wikipedia), while Canada, South Korea, and Israel are pursuing sector-specific, risk-based frameworks that remain largely non-binding (Wikipedia). Yet momentum toward global coordination is growing: at the 17th BRICS Summit in Rio, member nations issued a significant statement calling for UN-led multilateral AI governance focused on fostering innovation, transparency, privacy, and sustainability (BRICS, ECNS, Reuters, LinkedIn). India’s Prime Minister Modi also urged continued collaboration on responsible AI and securing supply chains for critical AI-related resources (Economic Times). While approaches differ across countries, the push for global regulation to balance innovation with ethics and security is gaining momentum.
As always, thank you for reading! Let me know what you found helpful, what you disagreed with — and what you’d like to see more of.
Many thanks,
Rebecca
