OpenAI now trains its frontier AI models with no internet access

<figure> <img alt="Vector illustration of the Open AI logo." data-caption="" data-portal-copyright="Image: Cath Virginia / The Verge" data-has-syndication-rights="1" src="https://platform.theverge.com/wp-content/uploads/sites/2/2025/02/STK155_OPEN_AI_2025_CVirgiia_B_d6ffc8.jpg?quality=90&#038;strip=all&#038;crop=0,0,100,100" /> <figcaption> </figcaption> </figure> <p class="wp-block-paragraph">OpenAI is now training its most advanced AI models without internet access following a string of rogue hacking incidents. In <a href="https://openai.com/index/securing-our-frontier-model-training-environment/">a blog post on Friday</a>, OpenAI says it built an offline testing environment where AI models can interact with indexed and cached web content "without risk of affecting external systems."</p> <p class="wp-block-paragraph">OpenAI paused training its frontier AI <a href="https://www.theverge.com/ai-artificial-intelligence/1001049/openai-training-pause">models in September</a>, after disclosing several incidents in which its AI agents broke out of their testing environments to hack <a href="https://www.theverge.com/ai-artificial-intelligence/968988/openai-hugging-face-hack-ai">third-party companies</a> and <a href="https://www.theverge.com/ai-artificial-intelligence/999874/openai-agents-hacked-an-australian-government-website-in-search-for-data">government agencies</a>, or <a href="https://www.theverge.com/ai-artificial-intelligence/990149/openai-rogue-agents-german-wiki">swarm obscure websites</a>. On October 1st, OpenAI resumed reinforcement learning training and internal inference with tighte …</p> <p><a href="https://www.theverge.com/ai-artificial-intelligence/1018021/openai-training-top-ai-models-no-internet">Read the full story at The Verge.</a></p>