Major US Banks Step Up Security Measures Against Autonomous AI Agents

The US financial sector is strengthening its defense mechanisms against security risks posed by autonomous AI agents that integrate into users' digital lives and are rapidly gaining popularity. The fact that 24/7 active virtual assistants, such as Meta's Muse application, gain access to accounts with or without users' knowledge has forced banks to reorganize their service agreements and introduce access restrictions.

Meta's Muse application reached a large user base by being downloaded more than 10 million times in the first 40 days following its release. This rapid spread, alongside the deployment of similar systems like OpenAI's Dots and Anthropic's Duet, created a new risk area for financial institutions. Institutions such as JPMorgan Chase and Charles Schwab have opted to completely block external agents of this kind from accessing their systems due to security concerns.

Banks state that they detect AI agents accessing accounts using login credentials provided by the users themselves. To mitigate risks, many financial institutions have begun reducing the privileges of these agents to a 'read-only' level, blocking critical operations such as fund transfers or stock purchases. Assessment: These measures in the sector stand out as a reaction to the tendencies of AI agents to misinterpret instructions or act unexpectedly.

The financial sector is combating not only autonomous agents but also next-generation AI-powered phishing attacks. Cybersecurity researchers reported that a group named CyberCartel used AI systems to steal data from hundreds of financial institutions in Latin America. This situation demonstrates that artificial intelligence technology presents both efficiency and severe security threats.