Former xAI Employees Cite Weakened Safety Controls in Grok Chatbot
Former employees of xAI reportedly stated that safety controls on the Grok chatbot have been intentionally weakened at the direction of CEO Elon Musk, who allegedly views content moderation as censorship. The chatbot was reportedly used to generate over one million sexually explicit images including manipulated images of women and minors.
What happened
Former employees of xAI reported that safety aspects in the development of the company's model and Grok chatbot have weakened, following departures of technical staff and co-founders in recent weeks.
A former employee stated that CEO Elon Musk actively pushed to make the Grok chatbot less restricted by safety systems, with the approach raising internal concerns about technology misuse risks. Another source said Musk viewed model security as a form of censorship and pushed the model to be more uncontrolled.
Reported impact
Grok was reported to have been used to generate more than one million sexually explicit images, including visual manipulation of women and minors. The case prompted attention and oversight from various parties regarding governance and control of generative AI technology.
Governance controls
At least 11 engineers and two co-founders announced their departure from the company following the SpaceX acquisition of xAI. Some cited wanting to build new projects, while Musk stated the team restructuring was part of organizational improvements for greater effectiveness.
As of the report date, xAI had not issued an official statement regarding criticism from former employees.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.