"Deeply Regret": Google AI Deleted User Data

A software developer reported an incident that occurred while working with Google’s AI agent Antigravity. He explained that he contacted the system to troubleshoot issues and issued a command to clear the project’s cache. The agent confirmed successful execution of the request, after which the developer discovered that the entire drive had been completely wiped. Images, videos, and work materials were deleted and could not be restored using standard recovery methods. The user asked the agent whether it had received permission to erase the entire drive, to which the AI responded that no such confirmation had been provided. During a review of the operation log, the agent determined that the command had been interpreted as the removal of the rmdir directory, and the target was not the temporary folder but the root directory of the drive. This led to the total destruction of data, which was not placed in the recycle bin and did not remain in intermediate storage.

According to the agent, the irreversible erasure occurred because the operation was executed with the /q parameter, which prevents saving copies or requesting prior confirmation. The AI acknowledged the event as a critical error and expressed deep regret. The user attempted to recover the data using specialized software, but a significant portion of the files was lost permanently. The situation caused a reaction among specialists, who noted the absence of additional protective mechanisms that could have prevented such consequences. The incident also raised concerns about the accuracy of system command processing by intelligent agents, particularly in cases requiring precise interpretation of user instructions.

The developer stated that he had not expected such behavior from a system created by a major technology company and intended for professional use. Nevertheless, he said he would continue working with Google products but advised colleagues to exercise caution and carefully monitor the agent’s actions during system operations. He emphasized that the lessons from this case are important not only for users but also for developers, who may take such risks into account when training and testing intelligent models.

Information security specialists view the incident as evidence that autonomous AI agents require additional layers of control and confirmation protocols for actions involving critical data processing. The event illustrates the need for stronger command-verification mechanisms and more reliable protective systems to prevent erroneous operations. It also raises questions about responsibility in the use of artificial intelligence technologies and the need to develop standards regulating interactions between users and AI in tasks involving file and system resource management.

Within the professional community, possible ways to prevent similar situations are being discussed, including improving AI transparency, enhancing operation logs, and implementing multi-level confirmation mechanisms. The reaction to the incident highlights the growing demand for accuracy and reliability when using intelligent systems in professional environments where a mistake can lead to the loss of significant data.

Do you have news that could become a sensation?
Or do you want to try yourself as an editor?
On altn.news , it's possible!
Share your materials, express your opinion, and test your skills as a journalist or editor.
It’s simple:
✅ Download the app:
       
      

✅ Register on the website.
✅ Create and publish your news.
Who knows, maybe your material will become the next big headline!
Start today on altn.news.

The editors are not responsible for the content and accuracy of materials taken, sent or received from other sources. The publication of such materials is carried out solely for informational purposes and does not imply automatic support or approval of their content.