Researchers use AI to find hidden loopholes in German law

Researchers say they used AI to find a substantial number of loopholes in German law, and that to their knowledge none had been recorded before. Security expert Bruce Schneier shared the work on his blog on November 3, 2026. He has written for years that AI could one day "hack" legal systems the way hackers break into computers.

What the researchers say

The team argues that written law works a bit like computer code. Both are written down, and both can contain flaws that someone clever can exploit. They used large language models, the kind of AI behind popular chatbots, to hunt for those flaws.

Courts usually absorb such flaws by interpreting what a law was meant to say. That is why most loopholes rarely come to light. The researchers believe cheap AI could change that by finding them at a scale not possible before.

They published only a selection, along with possible ways to exploit each one. They chose loopholes they believe would not cause serious harm if used.

What is not known yet

The material does not say how many loopholes were found, which laws are affected, or how the AI was tested. It also does not say whether anyone has actually used one. Schneier calls the work early, but says it is something coming.

The researchers suggest lawmakers would need to harden how laws are written, possibly by using advanced AI themselves.