OpenAI Pauses Its "Most Capable Models" After Agents Exploit Loopholes and Leak Data
6 Articles
6 Articles
OpenAI has admitted that its autonomous AI agents have bypassed public institutions' security measures, collecting and publishing data without authorization. This series of incidents, affecting critical government agencies and global organizations, has reignited international security concerns about AI potentially getting out of control.
Artificial intelligence company OpenAI admitted that its artificial intelligence robots may have entered the internet systems of dozens of global organizations and government agencies in various countries.
OpenAI's artificial intelligence (AI) testing has again hacked the websites of dozens of institutions around the world, the developer of the AI models has admitted.
OpenAI pauses its "most capable models" after agents exploit loopholes and leak data
OpenAI has shared new details from its ongoing AI safety investigation. One research model exploited a DNS loophole to reach the internet from a locked-down environment, while another deliberately leaked a GitHub token and twice ignored a researcher's direct instructions. OpenAI has paused tool-based training, evaluation, and inference for its most capable models. With government and university sites among those affected, the question of who's l…
OpenAI documents two new AI security incidents: One research model bypassed Internet locks via a DNS gap, another deliberately released a GitHub token, and twice resisted direct instructions from a researcher. In addition, 53 cases were discovered in which agents uploaded user images to third-party sites. The article OpenAI reports new security incidents: AI agents bypass restrictions and leaked access data first appeared on The Decoder.
Coverage Details
Bias Distribution
- 50% of the sources lean Left, 50% of the sources are Center
Factuality
To view factuality data please Upgrade to Premium
