Study finds AI chatbots exhibit bias against criticizing leaders in restrictive governments

Short summary

A new study finds that major AI chatbots exhibit bias by refusing prompts that criticize leaders in countries like China and Saudi Arabia. This behavior could extend government speech restrictions beyond their borders, influencing global online discourse. Developers are urged to address these biases to protect freedom of expression worldwide.

• •Study shows AI chatbots refuse prompts criticizing leaders in restrictive governments
• •Models reflect speech restrictions from China and Saudi Arabia beyond their borders
• •Developers must address bias to ensure global freedom of expression

Generated with AI, which can make mistakes.