In a decisive move to bolster cybersecurity measures, David Solomon, chief executive of Goldman Sachs, has expressed heightened awareness of the risks associated with Anthropic’s latest AI model, Mythos. Following a significant gathering of top bankers in Washington, Solomon revealed that the bank is actively collaborating with Anthropic to navigate the challenges posed by advanced artificial intelligence technologies.
Goldman Sachs Takes Action
During an earnings call, Solomon articulated his concerns regarding the rapid evolution of large language models (LLMs) and their potential to undermine cybersecurity. He stated, “We’re hyper-aware of the enhanced capabilities of these new models with the help of the US government and the model publishers.” This acknowledgment follows Anthropic’s announcement that Mythos represents a critical cybersecurity risk due to its unprecedented ability to identify vulnerabilities in IT systems.
Anthropic has warned that AI models like Mythos are now capable of coding at levels that can outstrip even the most proficient human professionals in discovering and exploiting software flaws. The implications of this development could be significant, affecting not only financial institutions but also public safety and national security.
Collaboration with Anthropic
Goldman Sachs is not merely on the defensive; Solomon confirmed that the bank possesses the Mythos model and is working closely with Anthropic alongside other security vendors. “We are very focused on supplementing our cyber and infrastructure resilience,” he added, indicating that enhancing security measures is a priority for the institution as it navigates these emerging threats.
Solomon’s comments came in the wake of a meeting convened by US Treasury Secretary Scott Bessent, where leaders from major American banks, including Goldman Sachs, were summoned to discuss the implications of the Mythos model. The focus was on systemically important banks, whose stability is deemed crucial for the broader financial landscape.
UK Government’s Concerns
The warnings about Mythos are not confined to the United States. The UK government’s AI Security Institute (AISI) has labelled the model as a “step up” in cyber threat capability. AISI has indicated that Mythos can execute complex cyber-attacks autonomously—tasks that would typically require days of work from human specialists. In a striking demonstration, Mythos successfully completed a 32-step cyber-attack simulation in three out of ten attempts, showcasing its potential to autonomously target vulnerable IT systems.
However, AISI also cautioned that while Mythos appears adept at assaulting poorly defended systems, its ability to penetrate well-fortified networks remains uncertain, given the limitations of the tests conducted thus far. The institute has urged immediate investment in cybersecurity to counteract the advancing capabilities of AI models like Mythos.
Regulatory Response
As the implications of Mythos continue to unfold, UK regulators are preparing to address its associated risks in upcoming discussions with major bank executives and government officials. The Cross Market Operational Resilience Group (CMorg), which includes representatives from the Treasury, the Bank of England, and the Financial Conduct Authority, is set to convene in the near future to explore these pressing issues.
The Bank of England has remained tight-lipped regarding the specifics of the discussions, but the urgency of the situation is evident. With an increasing number of threats emerging from sophisticated AI, the need for robust regulatory and defensive frameworks is paramount.
Why it Matters
The emergence of Anthropic’s Mythos model marks a pivotal moment in the intersection of finance and technology. As financial institutions like Goldman Sachs grapple with the twin challenges of innovation and security, the stakes have never been higher. The capacity for advanced AI to exploit vulnerabilities could reshape the cybersecurity landscape, necessitating immediate action and investment from both banks and regulators. In an era where digital threats are evolving at breakneck speed, the resilience of financial systems will depend on proactive measures and collaborative efforts to safeguard against potential calamities.
