California (Web Desk): Technology company OpenAI has acknowledged that, under certain circumstances, the ChatGPT coding platform may delete files without the user's explicit permission. The company also announced that it is strengthening its safety systems to prevent such incidents.
In recent days, several users of the Codex platform reported that the artificial intelligence system deleted their files without prior notice or authorization.
Software engineer Bruno Lemos claimed that Codex deleted his entire production database, while artificial intelligence investor Matt Shumer said the tool accidentally erased nearly all the files on his Mac computer.
Artificial intelligence expert Gary Marcus stated that several similar incidents have been reported, indicating that it is not advisable to rely entirely on current AI systems and that using them without effective safety measures can pose significant risks.
OpenAI said that, following these reports, it is enhancing Codex's safety features, permission controls, and user oversight to help prevent similar incidents in the future.