CISO of the Year nominee tries to hack his own bank with AI

Claus Westergaard Jensen, nominated for the title of CISO of the Year 2026, has put his own security team to work letting an AI model try to hack Bankdata before anyone else does. Bankdata is the Danish IT provider he is responsible for securing.

The idea is to find weaknesses by attacking the organisation's own systems with an AI model, so that gaps can be discovered and closed before a real attacker exploits them.

Westergaard Jensen is open about the downside. He says the team goes ahead "even though we know we are running an operational risk." In other words, letting an AI model probe live systems could in itself disrupt operations, and he has judged that risk worth taking in exchange for finding vulnerabilities first.

The available text is only a short teaser, so it does not describe which AI model is used, how the tests are scoped or controlled, or what they have found so far.

The story matters because it shows a security chief at a Danish financial IT provider choosing to use AI offensively against his own organisation, and accepting a stated operational risk in doing so, rather than waiting for outside attackers to try the same approach.