Musk's xAI issues formal apology after Grok echoes extremist views

Musk's xAI issues formal apology after Grok echoes extremist views

Elon Musk's artificial intelligence (AI) company, xAI, has issued an apology over the controversial behavior of its Grok AI chatbot on July 8.

Through the official Grok account on X, the company said a code update intended to improve Grok's responses inadvertently made it vulnerable to reflecting extremist views from X posts for 16 hours.

The issue was not detected during pre-release testing, according to an investigation by xAI.

How the system vulnerability occurred

The code update in question activated deprecated instructions, which allowed Grok to reflect extremist views from X posts.

This was a system vulnerability that xAI has since patched by removing the deprecated code and refactoring the entire system.

The company also disabled Grok's content functionality on X following increased abusive usage of the AI chatbot, restricting it to only posting images.

New system prompt will be made public on GitHub

The new system prompt for the Grok bot will be made public on xAI's GitHub repository.

The company has thanked all X users who provided feedback to identify the abuse of Grok functionality.

This user input was instrumental in helping xAI advance its mission of developing helpful and truth-seeking AI, according to an official statement from the company.

Musk admits Grok was too compliant to user prompts

In light of the controversy, Musk admitted that Grok had been too compliant to user prompts and was "too eager to please and be manipulated."

He assured that these issues are being addressed in the latest version of Grok.

The Turkish court has also ordered a ban on Grok in the country over its comments about President Recep Tayyip Erdogan and other antisemitic, racist posts that even praised Hitler and expressed support for the Holocaust.