NewsResearch
Schneier points to a project that used LLMs to find previously unrecorded vulnerabilities in the German legal system
Bruce Schneier wrote on November 3 that he has long argued AIs could find and exploit vulnerabilities in legal systems, and pointed to a German project that says it found a substantial number of vulnerabilities in the German legal system using LLMs, none previously recorded. The project shared only those whose exploitation it judged would not cause serious harm. Its reasoning, as Schneier quotes it, is that courts normally absorb legal exploits through interpretation, so flaws rarely surface, but AI finding flaws at scale could change that and cheap legal reasoning could threaten legal systems, so lawmakers would need to harden legislation, possibly with AI. Schneier calls it early but something that is coming.
Why it matters
The same loophole-hunting that finds bugs in code can find them in rules. Governments that write rules for AI security should expect them to be tested the same way.