In an era where artificial intelligence is evolving at breakneck speed, Goldman Sachs’ chief executive, David Solomon, has expressed heightened awareness of the risks posed by advanced AI models like Anthropic’s Mythos. Following a recent call to Washington, Solomon is keen to collaborate closely with the tech firm to bolster the bank’s cybersecurity measures.
A Wake-Up Call from Washington
Last week, Solomon joined other prominent bankers in Washington at the behest of US Treasury Secretary Scott Bessent. The meeting was pivotal, focusing on the implications of the Mythos AI model, which Anthropic has described as a game-changer in terms of cybersecurity risks. Solomon made it clear that the rapid advancements in AI, particularly in large language models (LLMs), are being closely monitored by Goldman Sachs to safeguard against potential cyber threats.
“Obviously the LLMs are making rapid progress and we’re hyper-aware of the enhanced capabilities of these new models with the help of the US government and the model publishers,” Solomon stated during an earnings call. His acknowledgment of the threats posed by AI highlights a growing awareness within the financial sector about the need for advanced protective measures.
The Capabilities of Mythos
Anthropic’s Mythos model has raised alarms for its unprecedented ability to identify and exploit vulnerabilities in IT systems, surpassing even the most skilled human professionals in this regard. 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.” This capability represents a significant shift in the landscape of cyber threats, with potentially severe consequences for economies and national security.
As Solomon elaborated, “We have the model. We’re working closely with Anthropic and all of our security vendors to harness frontier capabilities wherever it’s possible. This will continue to be an important focus.” His remarks underscore the commitment of Goldman Sachs to enhance its cyber and infrastructure resilience through strategic investments.
The UK Government’s Response
Meanwhile, the UK’s AI Security Institute (AISI) has echoed these concerns, branding Mythos a considerable escalation in the cyber threat environment. According to the AISI, the model can carry out complex cyber-attacks with multiple steps, discovering weaknesses in systems autonomously—tasks that would typically take human professionals days to accomplish. Notably, Mythos successfully completed a demanding 32-step cyber-attack simulation in just three attempts, demonstrating its formidable capabilities.
The AISI has cautioned that while Mythos appears adept at breaching poorly defended systems, it remains uncertain how effective it would be against well-fortified networks. As the capabilities of AI continue to advance, the AISI emphasised the necessity for immediate investment in cybersecurity measures to counteract these evolving threats.
Future Implications for Financial Institutions
In the coming weeks, UK regulators plan to address the risks associated with Mythos in meetings with British bank executives and government officials. The Cross Market Operational Resilience Group (CMorg) is set to convene shortly, including key figures from the Treasury, Bank of England, Financial Conduct Authority, and the National Cyber Security Centre.
As financial institutions grapple with these emerging threats, the stakes have never been higher. The Bank of England is currently coordinating communications regarding this crucial dialogue, further emphasising the urgency of the situation.
Why it Matters
The implications of advanced AI models like Mythos on the financial sector cannot be overstated. As Solomon and his peers prepare to tackle these challenges head-on, the need for robust cybersecurity strategies becomes paramount. The intersection of AI and financial security will define the future, and proactive measures are essential to ensure economic stability and protect sensitive data. In a world where technological advancements are both a boon and a threat, vigilance and innovation will be the keys to safeguarding our digital landscapes.
