Elon Musk has introduced plans to make X’s whole software program code publicly accessible after the platform completes a overview of potential safety vulnerabilities, positioning transparency on the centre of its response to current privateness issues surrounding xAI’s coding assistant, Grok Construct.
The announcement follows claims by a safety researcher that Grok Construct might have transferred considerably extra data than required whereas dealing with software program improvement duties. The allegations have triggered issues amongst builders and enterprises over whether or not confidential code and credentials might have been uncovered in the course of the course of.
Elon Musk has mentioned X will make its whole codebase publicly out there as soon as an inner safety evaluation is accomplished, a transfer introduced amid rising scrutiny over the info dealing with practices of xAI’s coding assistant, Grok Construct.
In a publish on X, Musk mentioned the platform can be absolutely open sourced after safety vulnerabilities are reviewed. He added that unbiased specialists would even be invited to confirm that the publicly launched code matches the software program working on the platform, describing transparency as important to constructing person belief.
As soon as now we have accomplished our overview for safety vulnerabilities, we’ll make all the codebase of ???? open supply, with no exceptions.
Furthermore, we’ll invite third celebration reviewers to look at the system that’s operating to verify that the open supply code is what’s operating.…
— Elon Musk (@elonmusk) July 15, 2026
The announcement comes after a safety researcher alleged that Grok Construct was transferring considerably extra person information than mandatory whereas processing coding requests. In response to findings cited by Axios, one take a look at reportedly concerned the add of 5.1 gigabytes of knowledge for a activity requiring solely 192 kilobytes, elevating issues that the software might have transmitted delicate improvement property, together with supply code, API credentials and different proprietary data.
The allegations sparked debate amongst builders over the safety of AI-powered coding assistants, significantly for organisations working with confidential software program repositories. Safety specialists urged affected customers ought to think about rotating API keys, cloud credentials and database passwords as a precaution, since deleting uploaded data doesn’t essentially remove the danger of prior publicity.
The problem additionally drew consideration from OpenAI Chief Government Officer Sam Altman, who described the reported incident as regarding in a publish on X. In a separate publish, Altman mentioned the episode bolstered the significance of open-source testing frameworks.
xAI later responded by stating that enterprise clients utilizing zero-data-retention (ZDR) agreements would not have their code or hint information saved. The corporate mentioned builders also can disable information retention utilizing privateness controls out there by means of Grok Construct’s command-line interface, which removes beforehand synchronised data.
Musk additional mentioned that every one person information uploaded to the platform earlier than the reported modifications can be completely deleted as a precautionary measure.