Language Models Identify Ambiguities and Exploit Loopholes

Open in new window