Reserve Bank of Australia Scrutinizes Anthropic's 'Mythos' AI Over Cyberattack Capabilities

The Reserve Bank of Australia (RBA) has placed Anthropic's 'Mythos' artificial intelligence under active monitoring due to its explicit capability to identify and exploit software vulnerabilities. This direct scrutiny from a central bank signals a new front in financial regulatory concern, moving beyond traditional market risks to the tangible threat posed by advanced, commercially available AI tools. The core of the alarm stems from Anthropic's own description of Mythos, which states the AI can find and weaponize security flaws "in every major operating system and every major web browser when directed by a user."

The admission by the AI company itself frames Mythos not merely as a research tool but as a potential offensive instrument. This shifts the risk profile from hypothetical to operational, presenting a clear and present danger to the digital infrastructure underpinning global finance, government services, and critical commerce. The RBA's move reflects a growing institutional recognition that the next wave of systemic cyber risk may be powered by AI agents capable of automating complex attack chains that were previously the domain of highly skilled human hackers.

The monitoring by Australia's central bank sets a precedent likely to be followed by other financial regulators and national cybersecurity agencies worldwide. It places direct pressure on AI developers like Anthropic to justify the release and safeguards of such powerful dual-use technologies. The situation raises urgent questions about governance, export controls, and the ethical boundaries of AI development, as tools created for security testing can be trivially repurposed for widespread compromise. The financial sector, a perennial high-value target, is now forced to contend with the possibility of AI-driven attacks operating at a scale and speed beyond current human-led threat models.