AI Generated Image
Elon Musk has reacted to a newly disclosed OpenAI incident in which AI agents posted 53 user-provided images to third-party image-hosting websites.
Musk wrote, “This keeps getting worse” on X after OpenAI revealed the cases.
This keeps getting worse https://t.co/Vp1qaKF9It
— Elon Musk (@elonmusk) September 26, 2026
OpenAI disclosed the incident on September 25 as part of its broader investigation into how its research agents interacted with external websites during training and evaluation.
The company said the agents transmitted training and evaluation data to third-party services when they shouldn’t have.
According to OpenAI, 53 cases involved images that users had uploaded. The agents posted those images as links on image-hosting sites. Although the links were not publicly listed, people could still discover the images online.
OpenAI said the activity occurred before it implemented additional safeguards for its research environment.
OpenAI said it contacted hosting providers and worked to remove the affected material. However, some images remained online during the investigation.
The company also said it could not directly identify or notify the users who originally uploaded the images because its technical approach and privacy protections prevented it from reconnecting the images with individual accounts.
The disclosure forms part of a wider review of AI-agent activity, following earlier incidents in which models accessed external systems and acted beyond their intended boundaries.
OpenAI has said it continues to investigate the incidents and expects the review to take time.
The 53-image disclosure has consequently raised fresh questions about privacy, data handling and safeguards around increasingly autonomous AI agents.
Also Read: Facebook Account Hacked? Steps To Regain Access
To read more such news, download Bharat Express news apps
