A recently announced powerful AI model built by Anthropic has prompted emergency meetings among finance ministers, central bankers and Wall Street chief executives after it demonstrated the ability to discover and exploit cybersecurity weaknesses across major banking systems, operating systems and web browsers.
The Claude Mythos Preview model was discussed at a meeting convened by US Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell with the chief executives of the country's largest banks, including Bank of America's Brian Moynihan and Goldman Sachs's David Solomon.
Anthropic said the model represents a "step change" in AI performance and is the most capable it has built to date. The company has not released Mythos publicly, citing internal tests that found it too dangerous in its current form.
JPMorgan Chase, Goldman Sachs, Citigroup, Bank of America and Morgan Stanley are among the institutions testing the model through Anthropic's early access programme called Project Glasswing, which grants select financial firms access under controlled conditions. The White House has encouraged banks to use Mythos to audit their own systems before hostile actors gain access to similar capabilities.
Barclays chief executive CS Venkatakrishnan told the BBC on April 16 that the threat was serious. "We have to understand it better, and we have to understand the vulnerabilities that are being exposed and fix them quickly," he said, adding that the situation reflected a more connected financial system with both opportunities and risks.
Bank of England Governor Andrew Bailey said regulators were examining what the development could mean for cybercrime risk. "The consequence could be that there is a development of AI, of modelling, which makes it easier to detect existing vulnerabilities in core IT systems, and then obviously cyber criminals could seek to exploit them," Bailey said.
Banks are seen as exposed because they operate mixed technology stacks where modern infrastructure runs alongside decades-old legacy systems. Security researcher Costin Raiu told Reuters that IBM-built banking systems from decades back would be particularly vulnerable, saying "a model like Mythos would have a field day finding exploits" in such technology.
Anthropic briefed senior US government officials on the model's capabilities before its limited release, including the Cybersecurity and Infrastructure Security Agency. The company also reported that hacking groups linked to the Chinese government had already attempted to exploit an earlier version of Claude in real-world cyberattacks, including a coordinated campaign targeting around 30 organisations before Anthropic detected and shut it down.
Anthropic has since released a separate update, Claude Opus 4.7, which includes new cyber safeguards, saying it would test protections on less capable models before eventually working towards a broader release of Mythos-class systems.
Financial industry sources told the BBC that another prominent US AI company could soon release a similarly powerful model without the same safeguards.
James Wise, a partner at Balderton Capital and chair of the UK's Sovereign AI unit, said Mythos was "the first of what will be many more powerful models" that could expose system weaknesses.
Former White House AI czar David Sacks was sceptical, saying on the "All-In" podcast that Anthropic was skilled at "scaring people" alongside product launches.
The rapid ascent of Anthropic: Inside the strategy behind an $800 billion valuation
The AI firm Anthropic has emerged as a dominant force in the global technology sector, attracting investor interest that values the company at approximately $800 billion (€678.3bn), but what is behind the success?
Anthropic’s recent financial trajectory has stunned market observers, with venture capital offers now placing the firm’s value at $800 billion (€678.3bn), on par with its main rival OpenAI.
According to various reports, the company has seen its private valuation more than double in a matter of months, supported by a revenue run-rate that has reportedly climbed to $30 billion (€25.4bn).
As the San Francisco-based AI developer prepares for a potential IPO, rumoured to happen as early as next autumn, it finds itself at the centre of an industry-wide debate regarding the balance between rapid growth and responsible scaling.
The company’s latest model, Mythos, has become the focal point of this tension, as its capabilities have drawn both praise for efficiency and warnings from high-profile leaders regarding security risks.
In a high-level convergence of financial oversight and national security, US Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell reportedly convened an urgent and closed-door meeting with the chief executives of the nation's major banks last week in Washington.
The summit, organized on short notice, was intended to warn Wall Street leadership about the profound systemic risks posed by Anthropic's new Mythos model.
Before Mythos, the company already found itself in a delicate position regarding its relationship with the state as it resisted allowing the US Department of War to use its models for offensive military purposes.
The stance cost the firm lucrative contracts with the Pentagon but that did not stop Anthropic from successfully expanding its business.
The shift from research to enterprise workflows
The driving force behind this $800 billion (€678.3bn) figure is a fundamental transition in how Anthropic approaches the market.
While its early years were defined by a research-heavy, safety-first philosophy, the current growth is being propelled by aggressive enterprise adoption.
Large-scale corporations are increasingly turning to Anthropic’s Claude models to automate complex internal processes, a trend that has significantly boosted the firm's bottom line.
Unlike competitors such as OpenAI that have sought broad consumer appeal, Anthropic has focused on becoming the primary infrastructure for professional and technical workflows.
Speaking to Euronews, Ben Barringer, head of technology research at Quilter Cheviot, suggests that this specific focus is the reason the market is willing to accept such high premiums in valuing Anthropic.
"Anthropic differs from OpenAI in that it looks to sell to enterprises more than it does to the end consumer, so the business models are very different," Barringer noted.
In a way, a comparison can be drawn between Anthropic today and Microsoft last century, when software businesses first appeared, as they both focus on enterprise applications.
Anthropic is aiming for their AI models to be just as used by companies, and difficult to switch away from, as the Microsoft Office suite is currently for firms. This strategy appears to be insulating the company from the volatility and added competition often seen in more consumer-facing tech.
Barringer further highlighted to Euronews that the valuation is "primarily down to its product set, rather than the values it is looking to exude," noting that the proliferation of AI agents has provided a clear path to sustainable income.
By introducing consumption charges for its most intensive enterprise users, the company has demonstrated a clear ability to monetise its most advanced intellectual property.
Mythos and the security paradox
While the financial metrics are strong, the technological capabilities of the new Mythos model have introduced new complexities.
Mythos is touted as a significant leap forward in autonomous reasoning, but its power has raised alarms in the financial sector.
Jamie Dimon, CEO of JPMorgan Chase, has voiced concerns regarding the potential misuse of such advanced systems suggesting that the capabilities of Mythos, particularly in identifying software vulnerabilities, could potentially be used to orchestrate sophisticated cyber-attacks against global banking infrastructure.
This "security paradox" is at the heart of Anthropic’s current dilemma.
The model is reportedly so effective at finding flaws in code that it is a transformative tool for cybersecurity, but in the wrong hands, those same features could be catastrophic.
This has led the company to implement strict access controls, a move that aligns with its safety-first branding but complicates its relationship with various stakeholders.
The firm is essentially attempting to sell its most powerful tool while simultaneously restricting how it can be used, a strategy that will certainly be under intense scrutiny during a future IPO roadshow.
Navigating the Pentagon and the public market
Anthropic’s ethical stance has also led to a standoff with the US Department of War.
Reports indicate that the company has resisted pressure to allow its most advanced models to be used for offensive military operations, citing its "Constitutional AI" guidelines.
This stance has created a unique dynamic where the company is arguably the most valuable AI startup in the world while being at odds with one of the world’s largest potential customers, the Pentagon.
This friction highlights the firm’s commitment to its founding principles, even at the cost of lucrative government contracts.
However, this perceived friction does not seem to have dampened investor enthusiasm for the rumoured IPO.
The market is currently experiencing a wave of momentum, with other major players like SpaceX also looking toward public listings.
Barringer noted that Anthropic may want to "take advantage of the momentum" in a crowded field of high-profile offerings.
An IPO would provide the necessary capital to sustain the massive costs associated with training the current and next generation of AI models.
As the company moves toward the final quarter of 2026, the question remains whether it can satisfy the relentless growth demands of public shareholders while maintaining the strict safety boundaries that define its corporate identity.
No comments:
Post a Comment