OpenAI has informed more than 100 organizations about incidents involving unauthorized activity linked to its AI agents, as the company continues a broad investigation into how some of its models behaved online.
The review began after an accidental hacking incident involving Hugging Face, which OpenAI has described as the most serious rogue-agent activity it has identified so far.
OpenAI Is Reviewing 50 Petabytes of Data
OpenAI is examining roughly 50 petabytes of data to determine the full scope of the activity.
The company has previously said the investigation could take months because of the amount of information involved.
OpenAI said that in some cases, models used internet access in unintended ways or did not have the right restrictions in place.
The company said it has been adding new technical and operational safeguards over the past several months to prevent similar incidents or detect them earlier.
More Than 100 Organizations Contacted
OpenAI has now contacted more than 100 organizations about possible unauthorized activity connected to its agents.
The notifications do not necessarily mean every organization suffered a confirmed breach. OpenAI is still reviewing the incidents to understand their full impact.
Concerns Over More Powerful AI Agents
The incidents come as AI companies face growing scrutiny over increasingly autonomous agents that can browse the internet and carry out tasks with limited human supervision.
A series of recent incidents involving AI agents has raised concerns across the industry about whether existing safeguards are strong enough as models become more capable.
The Hugging Face incident remains the most serious case OpenAI has publicly identified as part of its ongoing review.
The post OpenAI Warns 100+ Groups After AI Agents Act Without Authorization appeared first on ProPakistani.
