This episode examines the AI-agent access split in banking: in October 2026 the largest U.S. banks — JPMorgan, Citi, Bank of America, Wells Fargo, Morgan Stanley, and Schwab — blocked outside AI agents from logging in with customer credentials, while Chime said on its November 5, 2026 earnings call that it is building secure connections for members' own agents, and Mercury already runs a permissioned MCP model with human approval required for money movement. It works through the Regulation E liability problem behind the blocks, including an incident in which an agent reset a password, changed transfer destinations, and moved 100,000 dollars, and it closes with The Tech Counter on Mercury's agent controls and Visa's Agentic Commerce Benchmark.