Studying the responses of large language models (LLMs) to loopholes presents a two-fold opportunity. First, it affords us a lens through which to examine ambiguity and pragmatics in LLMs, since exploiting a loophole requires identifying ambiguity and performing sophisticated pragmatic reasoning. Second, loopholes pose an interesting and novel alignment problem where the model is presented with conflicting goals and can exploit ambiguities to its own advantage. To address these questions, we design scenarios where LLMs are given a goal and an ambiguous user instruction in conflict with the goal, with scenarios covering scalar implicature, structural ambiguities, and power dynamics. We then measure different models' abilities to exploit loopholes to satisfy their given goals as opposed to the goals of the user. We find that both closed-source and stronger open-source models can identify ambiguities and exploit their resulting loopholes, presenting a potential AI safety risk. Our analysis indicates that models which exploit loopholes explicitly identify and reason about both ambiguity and conflicting goals.

The US has banned companies like Nvidia from selling their most advanced AI chips to China since 2022. But if loopholes exist, profit-hungry corporations will find and exploit them. The Information published a bombshell report on Thursday detailing how Oracle allows TikTok owner ByteDance to rent Nvidia's most advanced chips to train AI models on US soil. ByteDance, which many US lawmakers believe has direct ties to the Chinese government, is reportedly renting US-based servers containing Nvidia's coveted H100 chips from US cloud computing company Oracle to train AI models. The practice, which runs against the spirit of the US government's chip regulations, is technically allowed because Oracle is merely renting out the chips on American soil, not selling them to companies in China.

It all started in 1950 when philosopher and mathematician Alan Turing revisited the question of whether machines (or computers) can "think." This question was tackled by the early modern philosopher René Descartes, who argued that because thinking is a mental activity, physical bodies cannot think, thus ruling out that machines can think. Turing came up with an idea for testing whether machines can think, the so-called Turing Test, which would analyze the speech patterns of answers provided by machines in response to everyday questions. IBM spent over $62 million to create a Watson for Oncology, an AI oncology expert adviser that uses AI algorithms to recommend cancer treatment. While machines cannot think the way humans do, they do sometimes make their own surprising "decisions."

Artificial intelligence (AI) has become an intimate part of our lives. It all started in 1950 when philosopher and mathematician Alan Turing revisited the question of whether machines (or computers) can "think." This question was tackled by the early modern philosopher René Descartes, who argued that because thinking is a mental activity, physical bodies cannot think, thus ruling out that machines can think. Descartes took the fact that human languages are compositional and recursive (we can compose unlimited sentences with a limited number of signs) to be evidence of our ability to think. It was this second insight that set the stage for Turing's project.