technology

OpenAI Trains Advanced AI Models in Offline Environment to Enhance Security

23 Ekim 2026The Verge

OpenAI has started training its advanced AI models without internet access to prevent hacking incidents, creating a secure offline environment.
OpenAI has begun training its advanced AI models in an offline environment to mitigate risks associated with internet access. This decision follows several hacking incidents that raised concerns about the security of their systems.
The new setup allows models to interact with indexed and cached web content while ensuring external systems remain unaffected.
The move comes in response to increasing incidents of hacking that have highlighted the need for better security measures in AI development. OpenAI's initiative could set a precedent for other organizations in the industry, emphasizing the importance of safeguarding AI systems from external threats.
This shift to offline training reflects a growing awareness of the vulnerabilities associated with AI development. By isolating the training process, OpenAI aims to enhance the security and reliability of its models.

NewsAI özeti

“OpenAI built an offline testing environment where AI models can interact with indexed and cached web content.”

This article is for informational purposes only and does not constitute professional advice.

Tam teknik rapor ve canlı veriler için yayıncının web sitesini ziyaret edin.

Her yerde okuyun. iOS ve Android için ödüllü uygulamalarımızı indirin.