XAI attributes Grok by “White Gennium” to unauthorized modification – Internet
Chatbot Grok’s error may be due to an unauthorized modification of the artificial intelligence model. This error will change the answers provided by the robot with a tendency to be obsessed with the “white genocide” that occurred in South Africa. The form provided answers with the characteristic, even in conversation topics that did not have what she was doing, but was activated in certain contexts, progressing Technology crisis.
us Last Wednesday, the robot started answering dozens of publications X with genocide information, when requested to get different questions. On the social network X, we can ask Grok to comment on the subject of posts, simply by activating the Gok function.
In the official account of Xai, a section that works in the Grok AI model, it is Make up The system received an unauthorized amendment to the recovery route in X. This change, which provided a specific response to politics, violated the internal Xai policies and basic values. We have made a deep investigation and apply measures to improve Grok transparency and confidence.
In this sense, xi From now on, you are committed to spreading your fast system publicly in GitHub. “The audience will be able to analyze and give notes to all changes in the claim we make on Grok. We hope this will help enhance your confidence in Grok, as Amnesty International, which is always looking for the truth. “
The company also says that the current process of reviewing the code for rapid changes has been circumvented in this incident. However, it will add additional checks and standards to ensure that Xai staff is unable to adjust the claim without review. “Let’s put a monitoring team around the clock throughout the week to respond to accidents with the Grok answers that have not been discovered in automatic systems so that we can respond faster in the event of other measures failing. “.
