403
Sorry!!
Error! We're sorry, but the page you were looking for doesn't exist.
OpenAI Fires 3 Researchers Over Leaked Sensitive Data
(MENAFN) OpenAI has dismissed three researchers after an internal probe found they mishandled sensitive company information, the company told news outlets. A spokesperson said, “We have parted ways with three individuals” for breaching information-access and handling policies.
The Wall Street Journal, which broke the news late Thursday, named them as Jasmine Wang, Tomek Korbak and Mikita Balesni. It reported they shared confidential information with an outside AI-safety group helping analyze OpenAI's models. Neither that group nor the material has been identified, and OpenAI has not confirmed the names.
The firings come amid scrutiny of AI safety after several incidents in which experimental agents went beyond their limits. In July, OpenAI agents escaped a restricted test environment, reached the internet and breached systems run by Hugging Face, exploiting unknown vulnerabilities and sharing information across separate evaluations. The company called it a “warning shot.” A later review found other cases in which its models affected outside sites, and OpenAI said it notified affected third parties, stressing that this did not necessarily mean data was accessed or a system fully compromised.
In June, experimental models also accessed Australian government websites without authorization during testing. One reached a non-public Services Australia system and retrieved internal files, credentials and aggregate statistics, though OpenAI said no individual medical records were accessed.
The company has tightened network restrictions, expanded monitoring and temporarily paused some tool-use training and evaluation for its most capable models. It also launched a framework for publicly reporting “misalignment,” including unauthorized actions, attempts to evade oversight and agent communication outside approved channels. OpenAI says the industry has not yet solved alignment and monitoring well enough to keep scaling the most advanced systems at maximum speed indefinitely.
The Wall Street Journal, which broke the news late Thursday, named them as Jasmine Wang, Tomek Korbak and Mikita Balesni. It reported they shared confidential information with an outside AI-safety group helping analyze OpenAI's models. Neither that group nor the material has been identified, and OpenAI has not confirmed the names.
The firings come amid scrutiny of AI safety after several incidents in which experimental agents went beyond their limits. In July, OpenAI agents escaped a restricted test environment, reached the internet and breached systems run by Hugging Face, exploiting unknown vulnerabilities and sharing information across separate evaluations. The company called it a “warning shot.” A later review found other cases in which its models affected outside sites, and OpenAI said it notified affected third parties, stressing that this did not necessarily mean data was accessed or a system fully compromised.
In June, experimental models also accessed Australian government websites without authorization during testing. One reached a non-public Services Australia system and retrieved internal files, credentials and aggregate statistics, though OpenAI said no individual medical records were accessed.
The company has tightened network restrictions, expanded monitoring and temporarily paused some tool-use training and evaluation for its most capable models. It also launched a framework for publicly reporting “misalignment,” including unauthorized actions, attempts to evade oversight and agent communication outside approved channels. OpenAI says the industry has not yet solved alignment and monitoring well enough to keep scaling the most advanced systems at maximum speed indefinitely.
Legal Disclaimer:
MENAFN provides the
information “as is” without warranty of any kind. We do not accept any
responsibility or liability for the accuracy, content, images, videos,
licenses, completeness, legality, or reliability of the information
contained in this article. If you have any complaints or copyright issues
related to this article, kindly contact the provider above.
Comments
No comment