June 29th 2025; My Research Analyst is a bot; Agents face the graveyard, India gets in on the Act & more on data centers sparking climate concerns. Here’s what I learnt this week on AI in Trading
1. My Research Analyst is Now a Bot
The irony! As Europe moves to rebundle research, AI is disrupting the traditional sell-side research model further still. Perplexity’s tools automate extracting and summarizing data from primary sources like US EDGAR, undercutting traditional research offerings as clients gain direct, faster insights (Perplexity blog). To stay relevant, banks will pivot toward proprietary analysis and bespoke advisory. Major banks are reported as cutting upto a third of junior tasks in research and modelling, shifting future jobs toward AI proficiency and sector expertise (MarketWatch article).
2. Agents Face Graveyards Amid “AgentWashing”
The ISG 2025 Agentic AI Report warns that despite financial services leading in agentic AI, only 25% of agents run autonomously, with data quality and governance challenges hindering complex use cases like trading or post-trade automation. Gartner predicts over 40% of agentic AI projects will be cancelled by 2027 due to hype, unclear value, and risk issues, leaving many firms with abandoned “agent graveyards.” Both reports claim that multi-agent systems still hold promise for dynamic trading and intelligent order routing – but know your Agents as the risk for hype and therefore wasted investment is high. Further reading: ISG Report, Gartner Press Release.
3. Data Dilemmas: Rights, Use, and Regulation
Banks and hedge funds are rapidly training proprietary NLP models for news and sentiment, but face IP, privacy, and legal risks. A US ruling in Anthropic’s case held that training AI on copyrighted books may qualify as transformative fair use (Wired article), adding complexity for firms using financial texts. Meanwhile, Europe’s EU AI Act imposes strict new requirements for high-risk systems like trading algorithms, mandating explainability, audits, and CE marking by 2026, with enforcement starting in just a few weeks in August 2025 (EU Digital Strategy, Transcend summary) including penalties, national competent authorities, and governance obligations making explainability a legal requirement rather than merely a best practice for firms deploying AI in sensitive domains like trading (Transcend timeline).
4. India’s SEBI Eyes AI Guardrails in Trading
India’s SEBI has proposed guidelines for responsible AI/ML use in securities markets to protect investors and market integrity (SEBI paper). Requirements include robust model governance, transparent AI disclosures, fairness and bias controls, pre-deployment testing, and strong data privacy protections. Firms have until July 11 2025 to submit feedback. The move underscores growing global momentum for AI regulation in financial markets.
5. AI’s Carbon Footprint Hits Trading Infrastructure
This week Sustainable Trading held its first meeting on the Carbon Footprint of a Trade (https://sustainable-trading.org). Despite claims of minimal impact, AI’s environmental toll is growing with Google announcing this week it’s emissions are up over 50% (https://www.theguardian.com/technology/2025/jun/27/google-emissions-ai-electricity-demand-derail-efforts-green). Financial markets, driven by vast data center infrastructure, are facing rising environmental footprints as trading volumes and regulatory data retention demands increase. EU sustainability rules are pushing firms to account for Scope 3 emissions across supply chains. While AI reduces manual processes, it deepens reliance on energy-hungry tech. Sustainable practices—like refurbished hardware and smarter cloud usage—are becoming essential for competitiveness and compliance. One to watch as collaboration across trading desks, vendors and regulators will be critical to build a greener trading ecosystem.
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
