Mythos AI model exposes banking and macOS security vulnerabilities
Anthropic's Mythos AI model can rapidly identify cybersecurity vulnerabilities in banking and macOS systems. Regulators including the European Central Bank and Dutch National Cyber Security Centre warned of potential misuse by criminals and recommended enhanced security measures.
What happened
Anthropic's Mythos AI model can rapidly identify cybersecurity vulnerabilities in banking systems and macOS. The model is currently available only to a select group of partners including technology companies, cybersecurity firms and banks.
Researchers have used Mythos to identify security flaws in macOS systems. The European Central Bank warned eurozone banks to prepare for potential cyberattacks using Mythos, describing the situation as urgent.
Reported impact
If criminals were to use Mythos, banking systems could become vulnerable to attack. The rapid identification of vulnerabilities by the AI model poses a concern for cybersecurity if deployed maliciously.
Index analysis
The European Central Bank called the situation urgent and instructed banks to identify and address possible weaknesses in their systems. The Dutch National Cyber Security Centre issued warnings with a series of recommendations.
Governance controls
Anthropic restricts Mythos access to select partners and requires them to test the model's capabilities and strengthen their own security based on results. Dutch banks report being in close contact with the National Cyber Security Centre and endorse its recommendations, with banks implementing measures to accelerate the time between vulnerability discovery and remediation.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.