ASIC Monitors Anthropic’s AI with Global Regulators
▼ Summary
– Australia’s ASIC is monitoring Anthropic’s Mythos AI model alongside international regulators due to its potential implications for financial markets.
– The Mythos model can identify and exploit zero-day vulnerabilities in major software, which regulators view as a potential systemic risk to financial infrastructure.
– Global financial regulators, including the Bank of England, the ECB, and the U.S. Federal Reserve, have responded with urgent, coordinated meetings to assess the cyber risks.
– Major banks and tech companies are partners in Anthropic’s restricted access program, using the model for defensive security testing while receiving significant credits from the company.
– Anthropic faces a separate legal challenge as the U.S. Department of Defense has designated it a supply chain risk, barring it from defense contracts.
The global financial regulatory community is now actively tracking the emergence of Anthropic’s Mythos AI model, with Australia’s securities watchdog joining a coordinated international effort. The Australian Securities and Investments Commission stated it is monitoring the model’s development alongside global peers to evaluate potential impacts on domestic markets. This announcement follows similar expressions of concern from major central banks and treasury departments, highlighting a unified regulatory focus on a single technological release.
ASIC emphasized that financial services licensees must proactively safeguard their clients in the face of such advanced AI systems. The regulator’s vigilance is part of a swift, worldwide reaction to Mythos, which Anthropic launched earlier this month under a controlled access initiative named Project Glasswing. The company claims the model can identify and exploit previously unknown software vulnerabilities across all major platforms, a capability intended for defensive security work but viewed by authorities as a potential systemic risk if obtained by malicious actors.
Financial regulators have responded with unusual speed and coordination. Bank of England Governor Andrew Bailey recently warned that Mythos could fundamentally reshape the cyber risk landscape, urging regulators to quickly assess its ability to find weaknesses in critical financial infrastructure. The bank’s resilience and AI groups have scheduled urgent meetings on the topic. European Central Bank President Christine Lagarde offered a candid assessment, noting that no existing governance framework is currently equipped to manage such powerful technology.
In the United States, Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell convened an emergency meeting with leading bank CEOs to discuss the implications. This session occurred while many executives were already in Washington for separate forum meetings, with JPMorgan Chase’s CEO as the sole notable absence. The Treasury Department confirmed the discussion and plans for ongoing dialogues with regulators and financial institutions.
Commercially, several major U. S. banks have begun internal evaluations of Mythos for defensive purposes. Goldman Sachs CEO David Solomon told analysts his bank has access and maintains hypersensitivity to the enhanced abilities of new AI. JPMorgan Chase is listed among the initial Project Glasswing partners, a group that includes roughly forty firms such as Amazon, Apple, Google, Microsoft, and Nvidia. Anthropic has committed substantial funding, including $100 million in credits for these partners and $4 million for open-source security groups, aiming to build defensive capacity before any wider release.
The primary concern for regulators is structural, not isolated. Financial institutions operate on complex technology stacks that blend outdated legacy systems with modern cloud infrastructure, creating significant technical debt and hidden vulnerabilities. The sector’s deep dependence on a handful of consolidated cloud providers means a capable AI exploiting flaws in those systems could trigger cascading failures across the entire financial network.
Anthropic’s restricted-access strategy has drawn criticism. IBM Senior Vice President Rob Thomas argued that security typically advances through transparency and scrutiny, not concealment. Anthropic CEO Dario Amodei defended the approach, stating that while the dangers of error are clear, success could enable a fundamentally more secure digital environment.
The company’s situation is further complicated by an ongoing dispute with the U. S. government. The Department of Defense has classified Anthropic as a supply chain risk to national security, a designation the firm is challenging in court. A federal appeals court recently denied Anthropic’s request to temporarily block the ruling, barring it from Defense contracts, though a separate injunction permits continued work with other government agencies as legal proceedings continue.
(Source: The Next Web)
