News
3 research pieces on “News” — quoted word-for-word from the filing, with a link back to the source
Bank of Thailand AI Risk Guidelines: 4 Rules for Using AI on 56-1
Bank of Thailand AI Risk Guidelines: 4 Rules for Using AI on 56-1
Retail investors drag a 56-1 into an LLM, ask whether the financials look good and whether to buy, get a polished summary that cannot point to its own sources, and trade on it. Meanwhile the Bank of Thailand has issued AI risk management guidelines covering both generative and agentic AI, anchored on the four FEAT principles, requiring a human in the loop wherever AI touches loan approval. This piece translates the regulator's standard into four hard rules a retail investor can apply immediately when talking to AI about their own money.
AI in Financial Services 2026: 80%+ Adopt, Only 40% See Profit
AI in Financial Services 2026: 80%+ Adopt, Only 40% See Profit
Cambridge's CCAF report on AI in financial services surveyed 628 organisations across 151 jurisdictions. Adoption runs above 80%, but 76% of large institutions cannot measure the value — the same failure as the retail investor who runs AI over financial statements every day and cannot say where the advantage is.
AI agents are moving real money — and fraud defences have not caught up
AI Agents Moving Money in Banks: Model Risk for Investors
June 2026: agentic payments crossed from demo to real money (Worldline-ING-Mastercard). Bank fraud defenses were built for humans, not AI — and that same governance gap is why an AI fabricates financial statement figures and leaves investors buying the top. The model risk view, and how I closed the gap in a working tool.