In an era where artificial intelligence is reshaping industries, Goldman Sachs is taking proactive measures to safeguard its operations against emerging cyber threats. CEO David Solomon has expressed an acute awareness of the capabilities of Anthropic’s latest AI model, Mythos, which has raised alarms about potential cybersecurity risks. The bank is collaborating closely with Anthropic to enhance its defenses as concerns mount over the model’s unprecedented ability to identify and exploit software vulnerabilities.
Heightened Awareness of AI Risks
During a recent earnings call, Solomon articulated the bank’s commitment to understanding the advancements in AI technology, particularly large language models (LLMs). “We’re hyper-aware of the enhanced capabilities of these new models with the help of the US government and the model publishers,” he noted, underscoring the urgency of adapting to the rapid evolution of AI.
The alarm bells were sounded by Anthropic last week when it revealed that Mythos could potentially expose significant flaws in IT systems. The company asserted in a blog post that AI models have reached a critical threshold in coding competency, capable of outperforming virtually all but the most skilled human experts in detecting and exploiting software vulnerabilities. The implications of this capability could be dire for economies, public safety, and national security.
Collaborative Efforts to Fortify Cyber Defenses
Solomon assured analysts that Goldman Sachs is taking these warnings seriously. “We’re working closely with Anthropic and all of our security vendors to harness frontier capabilities wherever possible,” he said. The bank is not only aware of Mythos but has also integrated the model into its security framework as part of a broader strategy to enhance its cyber resilience.
“Increasing our investment in cyber and infrastructure resilience is a top priority,” Solomon added, highlighting the bank’s proactive stance in the face of evolving technological threats.
The meeting last week in Washington, convened by US Treasury Secretary Scott Bessent, brought together Solomon and other leaders from major American banks. The discussion centred on the implications of the Mythos model, particularly for systemically important banks whose stability is crucial to the financial system.
UK Government Sounds the Alarm
The UK’s AI Security Institute (AISI) has also expressed grave concerns regarding Mythos. They describe it as a significant escalation in cyber risk compared to its predecessors. According to AISI, Mythos is capable of executing complex cyber-attacks autonomously, identifying system weaknesses without human input—tasks that would traditionally take days for skilled professionals to perform.
In a notable demonstration, Mythos became the first AI model to successfully navigate a 32-step cyber-attack simulation created by AISI, completing the challenge in three out of ten attempts. While AISI noted that Mythos seems adept at targeting less-secure systems, it remains uncertain about the model’s effectiveness against more robustly defended networks.
As the threats posed by advanced AI models like Mythos continue to evolve, AISI has urged immediate investment in cybersecurity measures to counteract these risks.
Upcoming Discussions Among UK Regulators
In the coming weeks, UK regulators are set to address the implications of Mythos with bank executives and government officials. The Cross Market Operational Resilience Group (CMorg), comprising chief executives and representatives from the Treasury, Bank of England, Financial Conduct Authority, and National Cyber Security Centre, will convene to discuss the pressing issues surrounding AI cybersecurity threats.
The Bank of England has opted to remain tight-lipped about the meeting, but the atmosphere is charged with urgency as financial institutions grapple with how to manage the risks posed by sophisticated AI technologies.
Why it Matters
As the intersection of finance and advanced technology deepens, the implications of AI models like Mythos extend far beyond the realm of banking. They pose existential threats to cybersecurity, economic stability, and even national security. Goldman Sachs’ proactive engagement with Anthropic highlights the necessity for financial institutions to stay ahead of the curve in technology adoption while fortifying their defenses against a landscape increasingly populated by intelligent, autonomous threats. The call to action is clear: investment in robust cybersecurity is not just prudent—it’s essential for safeguarding the future of financial operations in an AI-dominated world.