AI-augmented cyberattack compromises a G20 systemically important financial institution within 12 months
Following Bank of England Governor Bailey's formal G20 warning on AI threats to financial stability, at least one AI-augmented attack — deepfake-enabled fraud, automated social engineering, or AI-driven malware — is likely to compromise a G20 systemically important bank or financial market infrastructure within a year, prompting regulatory and industry response.
A confirmed US-Iran kinetic exchange — strikes on Larak Island, Iranian missiles on Gulf bases, and mines in the Strait of Hormuz — headlines today's outlook, with further US strikes more likely than not even as Trump's history of rapid face-saving de-escalation caps the risk. Secondary themes span accelerating NIMBY resistance to America's AI data-center buildout, South Asian and European fault lines over water and Ukraine policy, and a mounting AI-enabled threat to G20 financial institutions.
AI-augmented financial fraud is already operational reality (e.g., multimillion-dollar deepfake CEO/CFO fraud cases at major firms), and the threat surface is expanding rapidly. Over a full 365-day window, the probability that at least one G20 systemically important institution suffers a materially disruptive AI-augmented compromise is high; the main uncertainty is severity threshold and clean 'AI-enabled' attribution. My own cybersecurity track record shows an underprediction bias (~23pp under), which argues against anchoring at the analyst/Skeptic's 0.58. The cybersecurity analyst has no track-record data (default weight), so I lean on the strong external signal (Bailey's G20 warning) and the demonstrated base rate of AI-enabled bank fraud. I raise the estimate to 0.68, escaping the dead zone, while keeping it below high-conviction levels given attribution ambiguity.