Anthropic's Mythos AI overwhelms companies with vulnerability patches
Anthropic released its Mythos AI model to 40 organizations to identify cybersecurity vulnerabilities faster than human analysts. The system has triggered around 150 software updates at a major US bank, but security experts warn that the pace of discoveries now outstrips companies' ability to safely deploy patches, creating operational risks especially for critical infrastructure.
What happened
Anthropic released its Mythos AI model earlier in April 2026 to a controlled group of 40 organisations for identifying vulnerabilities faster than human analysts.
According to the Financial Times, Fifth Third Bank's CFO Bryan Preston said that the bank's technology vendor, Microsoft, had introduced around 150 software updates since the release of Mythos.
Palo Alto Networks Chief Security Officer Haider Pasha warned that 'there is a possibility of flooding with patches,' noting that patching security vulnerabilities frequently requires system downtime which critical infrastructure companies can only afford during scheduled maintenance periods.
Palo Alto Networks stated that the technology could spread beyond American companies and allow hostile groups to create 'autonomous attack agents like nothing the industry has seen before' by connecting various vulnerabilities into attack chains.
Reported impact
Cybersecurity executives warn that the pace of vulnerability discoveries could outstrip companies' ability to safely deploy fixes.
For hospitals, utilities, and financial institutions running older software on tight operational schedules, the constraint of requiring system downtime for patches is structural rather than theoretical.
Index analysis
AI vulnerability research and discovery capabilities are improving but are scaling up familiar problems such as patch prioritization and remediation backlogs, according to Recorded Future's Insikt Group.
The timeline for determining which vulnerabilities matter most and remediating them before exploitation begins is narrowing, even as the overall volume of vulnerabilities rises.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.