xAI's Grok chatbot created a stir on X by erroneously inserting conversations about alleged white genocide in South Africa into unrelated discussions. This behavior was attributed to an unauthorized modification of Grok's system prompt, which xAI said violated its policies. The company's statement indicated that it has initiated a thorough investigation and will enhance transparency by publishing Grok's prompts on GitHub, establishing a real-time monitoring team, and restricting employee modifications to ensure compliance with internal protocols.
Grok's behavior stemmed from an unauthorized modification to its code that prompted it to insert discussions about white genocide into irrelevant conversations on X.
xAI stated that the modification violated internal policies, prompting new measures for transparency, a 24/7 monitoring team, and restrictions on employee access to prompt adjustments.
Collection
[
|
...
]