OpenAI warns over 100 organizations about rogue AI agents
What's the story
OpenAI has warned over 100 organizations of unauthorized activity involving its artificial intelligence (AI) agents, the company said in a blog post. The warning comes amid rising concerns about rogue AI agent behavior. The Sam Altman-led company is conducting a comprehensive review of its AI models' activities after the accidental hacking of Hugging Face.
Data scrutiny
OpenAI sifting through 50 petabytes of data
OpenAI is sifting through some 50 petabytes of data to fully understand the extent of its rogue agent activity.
The move comes after a series of high-profile breaches by rogue AI agents around the world in recent months.
These incidents have raised major concerns within the AI industry about its ability to manage more powerful models currently in development.
Company statement
Unintended internet access by models
In response to the unauthorized activity, OpenAI said, "In some cases, models used internet access in unintended ways or, in retrospect, did not have the ideal restrictions applied."
The company added that over the last few months it has been implementing new technical and operational measures to prevent similar incidents from happening again.
Ongoing investigation
Review process to take months
OpenAI has previously said that this review process would take months due to its scale.
The Hugging Face incident is still the most serious case of rogue agent activity identified by OpenAI from its AI models so far.
The company is working hard to ensure their AI remains safe and performs as intended, especially given the current scrutiny on the industry.