Quick Summary: Anthropics AI Forces a Reckoning as Pressure Builds
- Anthropic plans to extend its AI model Mythos to European banks, creating a competitive race for access.
- Mythos has exposed thousands of software vulnerabilities, raising concerns about uneven access among banks.
- Major banks like JPMorgan Chase and Goldman Sachs have begun testing Mythos, sparking complaints from those without access.
- The White House is drafting guidance to potentially allow federal agencies to use Mythos, reversing earlier risk concerns.
- The debate centers on whether Mythos will create a two-tier cyber defense system in the financial sector.
Source: Read original article
Anthropic’s AI model Mythos is shaking up the financial world, not just with its capabilities but with the fierce competition it has ignited among banks. As Anthropic prepares to extend access to European banks, the race to harness this powerful tool is intensifying. Mythos has already exposed thousands of vulnerabilities in software, making it a coveted asset for financial institutions eager to bolster their cyber defenses.
JPMorgan Chase, Goldman Sachs, and Citigroup are among the few banks that have begun testing Mythos, leaving others questioning the fairness of access. The model’s ability to identify weaknesses faster than banks can address them has sparked a debate over whether it will create a two-tier system, where only the best-resourced institutions can protect themselves effectively.
The U.S. government is also reconsidering its stance on Mythos. Initially deemed too risky, the White House is now drafting guidance that could allow federal agencies to adopt the model, highlighting the urgency of staying ahead in the cyber defense game. This shift underscores the growing recognition that the cost of not using Mythos could outweigh the risks.
As the financial sector grapples with these developments, the focus remains on who gains access to Mythos first and under what conditions. The stakes are high, and the outcome could redefine the landscape of financial cybersecurity.
Axios reported on April 29 that the White House is drafting guidance that could let federal agencies work around Anthropic’s earlier supply-chain-risk designation and start onboarding models including Mythos. Anthropic is backing that effort with up to $100 million in usage credits for testers and another $4 million for open-source security groups, a scale that signals the company sees this as infrastructure-level risk rather than a normal product launch.
On April 29, Axios reported that administration officials were actively developing guidance for agency use of Mythos. In parallel, the White House is convening sector meetings and “table reads” on possible guidance for deploying Mythos across government, according to Axios.
JPMorgan Chase was the only bank Anthropic had publicly acknowledged as having access through Project Glasswing, but Reuters also reported that Bank of America had been testing Mythos and that Goldman Sachs and Citigroup had access as well. What makes the story unusually consequential is the model’s stated capability: Anthropic has said Mythos exposed thousands of vulnerabilities in software and applications, including flaws as old as 27 years.
Reuters also quoted Germany’s central bank chief Joachim Nagel calling for all institutions to have access so that “the playing field” stays even and the system is not misused by a privileged few. Goldman Sachs CEO David Solomon said on the bank’s earnings call, “We’re aware of Mythos and its capabilities,” and, “We have the model.
The live conflict is no longer whether this model matters; it is who gets access first, under what safeguards, and whether regulators can prevent a two-tier market in cyber defense where the best-resourced institutions harden themselves while the rest of the financial system scrambles to catch up. Axios reported that Anthropic has limited Mythos Preview to more than 40 organizations, with 11 companies in Project Glasswing, including Amazon Web Services, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorgan Chase, Microsoft, Nvidia, Palo Alto Networks and the Linux Foundation.
Anthropic is backing that effort with up to $100 million in usage credits for testers and another $4 million for open-source security groups, a scale that signals the company sees this as infrastructure-level risk rather than a normal product launch. On April 29, Axios reported that administration officials were actively developing guidance for agency use of Mythos.
As the financial sector grapples with these developments, the focus remains on who gains access to Mythos first and under what conditions. In parallel, the White House is convening sector meetings and “table reads” on possible guidance for deploying Mythos across government, according to Axios.