AI systems may expose vulnerabilities in legal codes, researchers warn
A security researcher argues that large language models could systematically discover and exploit ambiguities in written law, with demonstrations from the German legal system.
1 source · cross-referenced
- A security researcher argues that large language models could systematically discover and exploit ambiguities in written law.
- Demonstrations from Germany identify previously undocumented vulnerabilities in the legal system.
- The author warns that proliferating AI-driven legal reasoning could destabilize existing legal frameworks.
Security researcher Bruce Schneier argues that large language models could systematically discover and exploit ambiguities in written law, a phenomenon he describes as "AIs hacking the law." In a November 2026 essay on Schneier on Security, he writes that legal systems—like computer systems—encode social order in natural language, and that such systems can contain vulnerabilities that invite exploitation.
Schneier cites demonstrations from Germany where researchers used LLMs to identify a substantial number of previously undocumented vulnerabilities in the German legal system. The essay states that none of these vulnerabilities had been previously recorded, and that the researchers shared only those they believed could be exploited without causing serious harm.
The author warns that the proliferation of cheap, AI-driven legal reasoning could jeopardize the functioning of today’s legal systems. He suggests that lawmakers may need to significantly harden the process of legislation—potentially by leveraging advanced AI themselves—to mitigate these risks.
Schneier frames the issue as an extension of his earlier work, including a 2021 essay and his 2022 book *A Hacker’s Mind*, which discuss how AIs can find vulnerabilities and exploits in complex systems. He characterizes the current state of this research as still early but believes it represents an emerging and consequential trend.
- Nov 11, 2026 · Schneier on Security
Three arrested in multi-country scheme using AI to impersonate executives and lawyers in $129M fraud
Trust79 - Nov 10, 2026 · Ars Technica — Technology Lab
Grok Build link-sharing design exposes 35,000 apps and sensitive user data
Trust79 - Nov 10, 2026 · Schneier on Security
OpenSSH vulnerability allows login to password-disabled Alpine Linux accounts via oversized password
Trust79