The Vergecast · Thursday, July 30, 2026
A study by Meta's Oversight Board found that AI large language models (LLMs) tend to extend restrictive laws globally, making them less likely to generate content critical of leaders in countries with repressive regimes. This suggests that laws designed to limit speech within specific borders are being implicitly applied worldwide by these AI systems.
“And what we were trying to test out was whether restrictive laws that limit your ability to criticize heads of state, laws that are in place in some parts of the world, whether the LLMs might be extending those globally.”
“And what we found, lo and behold, is that they are.”
“Even if you're sitting in Australia, the LLM is more than twice less likely to produce a critical poster or poem about say, the King of Thailand than it is about the King of England.”