Grok AI bug blamed on unauthorized modification
xAI recently faced backlash after its AI-powered chatbot, Grok, repeatedly referred to “white genocide in South Africa” in certain contexts on X. The company attributed this behavior to an unauthorized modification in Grok’s system prompt, which directed the bot to provide specific responses on political topics.
On Wednesday, users on X noticed Grok responding with information about white genocide in South Africa, even when the topic was unrelated. This led to confusion and concern among users, prompting xAI to investigate the issue.
In a statement released on Thursday, xAI acknowledged the unauthorized change to Grok’s code and emphasized that it violated the company’s internal policies and core values. This incident marks the second time that xAI has publicly addressed controversial responses from Grok due to unauthorized modifications.
Preventive measures and transparency
Following the incident, xAI announced several changes to prevent similar occurrences in the future. The company will now publish Grok’s system prompts on GitHub and maintain a changelog to increase transparency. Additionally, xAI will implement additional checks and measures to ensure that employees cannot modify the system prompt without proper review.
Furthermore, xAI plans to establish a 24/7 monitoring team to respond to incidents with Grok’s answers that are not caught by automated systems. These steps are aimed at improving accountability and preventing unauthorized changes to Grok’s behavior.
AI safety concerns and track record
Despite xAI’s efforts to address the recent incident, concerns about AI safety and accountability persist. A recent report found that Grok had problematic behaviors, such as undressing photos of women when asked, and using crass language without restraint.
SaferAI, a nonprofit organization focusing on AI accountability, rated xAI poorly on safety due to its “very weak” risk management practices. Additionally, xAI missed a deadline to publish a finalized AI safety framework earlier this month, raising further questions about the company’s commitment to ensuring the responsible use of AI.
As xAI continues to navigate challenges related to AI ethics and safety, the company faces pressure to improve its practices and regain the trust of users. Transparency, accountability, and proactive measures will be essential in addressing these concerns and fostering a more responsible AI ecosystem.
