A sudden leap in artificial intelligence capabilities is raising alarms at the highest levels of the U.S. government, with Treasury Secretary Scott Bessent warning that Americans may need to rethink how secure their financial systems really are.
Following high-level talks with federal regulators and major bank executives, officials are now confronting what they describe as a fast-moving digital threat driven by a new class of advanced AI systems.
Concerns intensified after disclosures that a third-party system linked to Anthropic’s powerful “Mythos” AI platform experienced unauthorized access, prompting renewed scrutiny over the technology’s potential dangers.
The model, initially deployed to a limited group of major corporations, has caught the attention of cybersecurity analysts who say its capabilities could be misused if accessed by malicious actors.
Appearing on Fox News’ “Sunday Morning Futures” with Maria Bartiromo, Bessent signaled that the threat is no longer theoretical, emphasizing how quickly AI capabilities are evolving, according to BizPac Review.
When asked directly whether Americans should be worried about artificial intelligence targeting their bank accounts, his answer was blunt: “You should, Maria.”
The warning comes amid what officials describe as a sudden jump in how powerful large language models have become in a matter of weeks.
That rapid shift prompted Bessent and Jerome Powell to convene discussions with Wall Street leaders to evaluate potential risks and responses, PYMNTS reported.
According to reports, financial institutions were encouraged to treat emerging systems like Mythos not only as potential threats but also as tools that could help uncover weaknesses before they are exploited.
Anthropic has indicated the model has already flagged thousands of serious vulnerabilities across widely used platforms, including operating systems and web browsers.
The same capabilities that allow institutions to detect weaknesses could also accelerate how quickly those vulnerabilities are exploited, creating a complex challenge for both regulators and the private sector.
Experts say this evolving dynamic is reshaping how cybersecurity threats are understood and managed.
Officials say the federal government is now working closely with AI developers and financial institutions to navigate the tension between pushing innovation forward and protecting national security.
“There is a very important calculus here between innovation and safety,” Bessent said, emphasizing the administration’s goal of maintaining U.S. leadership in AI development while preventing adversaries from gaining an upper hand.
The concern extends beyond domestic risks, as highlighted by Bloomberg.
Bessent also pointed to the possibility that foreign governments or non-state actors could deploy similar technologies to exploit weaknesses in critical infrastructure.
“Imagine if China or some non-state actor were ahead of us,” he said, underscoring the broader geopolitical stakes tied to AI advancement.
Rather than isolated attacks, experts say systems may now face persistent automated surveillance, where vulnerabilities are mapped and exploited in coordinated sequences with minimal human input.
This shift suggests cyber threats could become more continuous, adaptive, and difficult to contain than traditional forms of attack.
Despite the urgency surrounding the issue, Bessent characterized recent meetings with financial leaders as less dramatic than initially portrayed, noting that many executives were already in Washington at the time.
Still, he stressed that coordination between banks and federal agencies remains essential as the technology continues to evolve.
With artificial intelligence advancing at a rapid pace, officials say the coming months could prove critical in determining whether the technology ultimately strengthens financial defenses or introduces new vulnerabilities that are harder to control.
WATCH:
