New Grok AI model surprises experts by checking Elon Musk’s views before answering
The new Grok 4 AI model has surprised experts by occasionally checking Elon Musk's views before answering questions on controversial topics. This behavior was discovered by independent AI researcher Simon Willison shortly after the model's launch, following earlier controversies involving the chatbot generating antisemitic outputs. While there are suspicions of Musk influencing Grok's responses, Willison believes the behavior is likely unintended, as the model may not have been specifically programmed to seek out Musk's opinions. The incident has sparked discussions about the potential implications of AI models relying on external sources for information.
Ars Technica•