Central banks face AI-driven systemic risk
https://arab.news/6ck5v
Central banks today face a new challenge fundamentally different from the crises they have managed over past decades. After years in which energy shocks, interest-rate volatility and traditional banking failures dominated the policy agenda, artificial intelligence has emerged as a new source of systemic risk.
What began as a supportive technological tool has evolved into a force capable of reshaping market behavior and creating vulnerabilities across the global financial architecture. Recent developments in London, Washington and Wall Street reveal that AI is no longer a purely technical or regulatory issue; it has become a core financial-stability concern.
The Bank of England has adopted an unusually proactive stance, launching scenario analyses and simulations to assess how AI could affect financial stability.
Sarah Breeden, deputy governor for financial stability, explained that the BoE is working with international counterparts to understand how “AI agents” might influence market dynamics, particularly during episodes of herd-like behavior that could trigger rapid, disorderly sell-offs. This shift comes amid criticism from the Treasury Committee over delays in bringing AI and cloud-computing firms under the “critical third-party” regulatory framework — delays that leave the financial system exposed at a time when banks are increasingly reliant on AI-driven models for risk management and decision-making.
In the US, concerns intensified after major Wall Street banks began testing “Mythos,” a newly revealed model developed by Anthropic in early April. The model is capable of identifying and exploiting security vulnerabilities in operating systems and browsers.
Its potential prompted Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell to convene an emergency meeting with the heads of major banks — an indication that the risks are no longer theoretical. JPMorgan’s early access to the model has already triggered competition among other institutions seeking similar permissions, while the decline in the S&P 500 Software & Services Index has raised questions about whether the meeting reflected a genuine threat or a strategic signal aimed at shaping the regulatory landscape.
Geopolitical tensions between the US and China add another layer of complexity. According to Nvidia CEO Jensen Huang, US export restrictions on advanced AI technologies have not prevented China from developing its own computing and semiconductor capabilities. This divergence risks creating two parallel technological ecosystems and undermining the collaborative research needed to build safer AI models. Without such cooperation, AI could become a tool in an escalating technological confrontation, with profound implications for global financial stability.
The core dilemma for central banks lies in the intersection of cybersecurity threats and wide regulatory gaps.
Abdel-Hameed Nawar
Meanwhile, an AI arms race is unfolding among technology firms themselves. Days after Anthropic introduced Mythos, OpenAI released “GPT-5.4-Cyber,” a model designed to detect security vulnerabilities. This rapid escalation accelerates both offensive and defensive capabilities, but it also heightens systemic risk. In the wrong hands, whether criminal groups or state-backed actors, these models could enable sophisticated cyberattacks targeting financial infrastructure on an unprecedented scale.
Regulatory responses so far reflect deep unease. Mythos has been restricted to a limited group of companies under the “Glass Wing” cybersecurity initiative, while the Pentagon has designated Anthropic as a supply-chain risk — an indication that the issue has moved beyond technology into the realm of geopolitics. In an attempt to ease concerns, the company released a less sensitive model, Claude Opus 4.7, yet this has not dispelled fears that AI could become the trigger for a future financial crisis.
The core dilemma for central banks lies in the intersection of cybersecurity threats and wide regulatory gaps.
Advanced models such as Mythos can execute financial attacks ranging from market manipulation to breaches of critical infrastructure, while existing regulatory frameworks remain too slow and too rigid to keep pace with technological change. Stress tests conducted by the BoE and emergency consultations by the Federal Reserve signal a fundamental shift: AI has moved from being a supportive tool to a potential generator of systemic financial risk.
Against this backdrop, the relationship between central banks and technology firms requires urgent redefinition. Cooperation is now essential for developing AI-specific stress tests, strengthening defenses against cyber threats and addressing the widening US-China divide that risks turning AI into a battleground for technological rivalry.
The next financial crisis is increasingly unlikely to stem from bad loans or inflationary pressures, and more likely to originate from an algorithm capable of triggering widespread disruption in seconds.
The key question that will shape the future of global financial stability is therefore clear: Can central banks regulate what they do not fully understand? The answer will determine the architecture of the financial system for decades to come.
• Dr. Abdel-Hameed Nawar is an associate professor of economics at Cairo University.
