Goldman Sachs is on high alert regarding the potential cybersecurity risks posed by Anthropic’s latest AI model, Mythos. The investment bank’s CEO, David Solomon, has voiced his concerns and announced a collaborative effort with the tech firm to enhance their cyber defences. As global attention on artificial intelligence grows, Solomon’s remarks underscore the pressing need for financial institutions to safeguard their operations against increasingly sophisticated attacks.
A Meeting in Washington
Last week, Solomon was among a select group of banking executives called to Washington for discussions centred around the implications of Mythos. The meeting, organised by US Treasury Secretary Scott Bessent, focused on systemically important banks—those whose disruption could threaten financial stability. This gathering highlighted the urgency with which regulators view the evolving landscape of AI and its potential to compromise security.
Understanding Mythos AI
Anthropic has raised alarms regarding Mythos, asserting that its capabilities could significantly outpace human experts in identifying and exploiting vulnerabilities within IT systems. In a recent blog post, the company noted, “AI models have reached a level of coding capability where they can surpass all but the most skilled humans at finding and exploiting software vulnerabilities.” With this in mind, Solomon remarked, “We’re aware of Mythos and its capabilities… We have the model. We’re working closely with Anthropic and all of our security vendors to harness frontier capabilities wherever possible.”
The advances represented by Mythos are not just theoretical; they present real threats that could have severe consequences for economies and public safety. Solomon acknowledged the need for Goldman Sachs to fortify its cybersecurity measures, stating, “We are very focused on supplementing our cyber and infrastructure resilience.”
The UK’s Response
In a parallel development, the UK government’s AI Security Institute (AISI) has issued warnings about Mythos, describing it as a significant escalation in terms of cyber threats. The AISI’s findings indicate that Mythos can execute complex attacks, discovering weaknesses in IT systems without human involvement—tasks that would typically require days for human professionals. In a striking demonstration, Mythos successfully completed a 32-step cyber-attack simulation, solving the challenge in three out of ten attempts.
As UK regulators prepare to address the risks posed by Mythos with British banking leaders and government officials, the urgency for enhanced cybersecurity measures cannot be overstated. The Cross Market Operational Resilience Group (CMorg), which includes chief executives and representatives from various financial and security sectors, is expected to convene within the fortnight to discuss these pressing concerns.
Investment in Cyber Defence
As the digital landscape becomes increasingly complex, the need for robust cyber defences is paramount. The AISI’s warning that future AI models will only improve upon Mythos serves as a clarion call for immediate investment in cybersecurity. Both US and UK financial institutions must act decisively to ensure they are not caught off-guard by the next wave of AI-driven threats.
Why it Matters
The implications of Solomon’s proactive approach towards the challenges posed by Mythos extend beyond Goldman Sachs. As one of the world’s leading financial institutions, Goldman’s actions could set a precedent for the entire banking sector on how to tackle emerging cybersecurity threats. In an age where AI capabilities are advancing at breakneck speed, a cooperative and forward-thinking strategy is essential for protecting not just individual banks, but the stability of global financial systems as a whole.
