OpenAI has updated its security protocols to combat corporate espionage, following the January release of DeepSeek, a Chinese startup accused of replicating OpenAI’s technology. The Financial Times reports that OpenAI’s enhanced security measures include “information tenting” policies, limiting employee access to sensitive project information, such as during the development of the o1 model. Only authorized team members could discuss the project in shared areas. Additionally, proprietary technologies are now isolated in offline systems, and biometric controls, including fingerprint scans, are required for certain office access. A “deny-by-default” internet policy ensures all external network connections need explicit approval. These changes aim to safeguard OpenAI’s intellectual property from foreign adversaries, while also addressing internal security concerns amid rising competition and data leaks within the American AI sector. OpenAI has not yet publicly commented on these revisions.
For the latest insights and analyses in tech, subscribe to our newsletter.
Source link