{"slug": "why-less-visibility-into-how-openais-new-gpt-6-astra-thinks-is-sparking-safety", "title": "Why less visibility into how OpenAI’s new GPT-6 Astra ‘thinks’ is sparking safety concerns", "summary": "OpenAI's new GPT-6 Astra model, announced Thursday, is sparking safety concerns because its written reasoning is 'harder to monitor' than the previous GPT-5.6 Sol, according to the company. The reduced visibility stems from a technique called recurrent depth, which processes logic in hidden mathematical loops rather than readable text, raising concerns among analysts about the ability to inspect model behavior, especially after the Hugging Face breach in July.", "body_md": "# Why less visibility into how OpenAI’s new GPT-6 Astra ‘thinks’ is sparking safety concerns\n\nThe Hugging Face breach in July highlights the importance of being able to inspect what models are ‘thinking’, say analysts\n\n[required a Chinese open model](https://www.scmp.com/tech/tech-trends/article/3361450/hugging-face-deploys-zhipus-glm-52-model-contain-autonomous-openai-cyberattack?module=inline&pgtype=article)to investigate, according to analysts.\n\nWhen announcing Astra on Thursday, OpenAI said it was “the world’s most intelligent and aligned model,” with a “significant jump in cyber capabilities”.\n\nOpenAI president Greg Brockman said at the end of a press call announcing Astra’s arrival that it likely represents AGI, or artificial general intelligence – AI that matches or outperforms human intelligence.\n\nHowever, OpenAI also said the model’s written reasoning was “harder to monitor” compared with GPT-5.6 Sol, the previous generation released in July.\n\n“We have found that GPT-6 Astra is more capable of controlling its own CoT (chain of thought) than GPT-5.6 Sol, and less likely to include incriminating information in its CoT,” OpenAI said, referring to the intermediate reasoning steps an AI generates while solving a task.\n\nThe shift in visibility stems from a technique known as recurrent depth, or looped transformers, which reuses parts of a neural network. As a result, it processes complex logic inside hidden mathematical loops rather than in step-by-step readable text, The Information reported on Tuesday ahead of the launch.", "url": "https://wpnews.pro/news/why-less-visibility-into-how-openais-new-gpt-6-astra-thinks-is-sparking-safety", "canonical_source": "https://www.scmp.com/tech/tech-trends/article/3366401/why-less-visibility-how-openais-new-gpt-6-astra-thinks-sparking-safety-concerns?utm_source=rss_feed", "published_at": "2026-09-04 10:31:21+00:00", "updated_at": "2026-09-04 10:54:21.589550+00:00", "lang": "en", "topics": ["artificial-intelligence", "large-language-models", "ai-safety"], "entities": ["OpenAI", "GPT-6 Astra", "GPT-5.6 Sol", "Greg Brockman", "Hugging Face"], "alternates": {"html": "https://wpnews.pro/news/why-less-visibility-into-how-openais-new-gpt-6-astra-thinks-is-sparking-safety", "markdown": "https://wpnews.pro/news/why-less-visibility-into-how-openais-new-gpt-6-astra-thinks-is-sparking-safety.md", "text": "https://wpnews.pro/news/why-less-visibility-into-how-openais-new-gpt-6-astra-thinks-is-sparking-safety.txt", "jsonld": "https://wpnews.pro/news/why-less-visibility-into-how-openais-new-gpt-6-astra-thinks-is-sparking-safety.jsonld"}}