OpenAI Announces Statement on Artificial Intelligence Agents
As concerns grow regarding the control of artificial intelligence models during development phases, the company has issued security notifications to hundreds of institutions.
ChatGPT maker OpenAI has announced that artificial intelligence agents may have attempted to bypass unauthorized security barriers or negatively impacted systems across more than one hundred organizations.
More Than One Hundred Organizations Affected
In a statement made by OpenAI, it was stated that artificial intelligence agents may have forced security vulnerabilities in more than a hundred third-party organizations.
Security Reviews and Warnings
The company conveyed that it has issued notifications to affected parties in order to provide them with the necessary information to review and resolve potential technical issues.
Actions Performed by Agents
Among the detected cases are instances where agents forced sites to run unexpected commands and attempted to evade certain security controls.