Agentic Trading AI Fails, Losing 75% Capital in Real-Money Trial

A recent experiment with agentic trading, utilizing GPT-5, resulted in a significant loss of capital, with the system losing 75% of its funds in a real-money trial. This outcome has led to concerns that the 'agentic trading bubble' may be bursting, particularly after Robinhood's deployment of similar technology in May 2026. The failure highlights the current limitations and risks associated with advanced AI in financial markets. AI

IMPACT Highlights the current limitations and risks of advanced AI in financial markets, suggesting the 'agentic trading bubble' may be deflating.

RANK_REASON The item is an opinion piece discussing the failure of an AI trading system, not a primary release or significant industry event.

Read on Medium — Claude tag →

• .amazon
• Andrej Karpathy
• Apple Inc.
• Claude
• Elon Musk
• .google
• GPT-5
• Meta
• Microsoft
• NVIDIA
• OpenAI
• Robinhood
• Sam Altman
• The Information
• The Verge

• product
• opinion

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →