OpenAI strengthens distillation safeguards in GPT-6.2 Astra and Sol
AI

OpenAI strengthens distillation safeguards in GPT-6.2 Astra and Sol

TechNews Editorial
TechNews EditorialNov 21, 2026 · 1 min read
Share

Why it matters

OpenAI says extracted reasoning can help train another model to reproduce capabilities without the safeguards that govern the original model.

The facts

  • OpenAI says it strengthened reasoning extraction safeguards in GPT-6.2 Astra and Sol.
  • In an internal test, average attack success fell from 54.0% on GPT-6 Astra to 0.2% on GPT-6.2 Astra.
  • OpenAI says it is applying selected safeguard improvements to GPT-6 Astra.

Advertisement

OpenAI says it has strengthened protections against attempts to extract private reasoning from GPT-6.2 Astra and GPT-6.2 Sol. It also outlined how it detects coordinated efforts to use its models’ outputs to train other models without authorization.

OpenAI calls this activity adversarial distillation. A model trained on another model’s answers or reasoning could gain useful capabilities without inheriting its safeguards, the company says. OpenAI is particularly concerned about cybersecurity and biology, where capabilities can have both beneficial and harmful uses.

Extraction attack success dropped below one percent

In an internal test using automated attackers, the average success rate for recovering reasoning generated during an earlier prompt fell from 54.0% against GPT-6 Astra with its original safeguards to 0.2% against GPT-6.2 Astra with updated safeguards. OpenAI says separate adaptive tests found success rates below 1% for two types of reasoning extraction attack against GPT-6.2 Astra.

The protections include model training, checks on incoming requests and outgoing responses, and API controls. OpenAI says its output checks cover tool-call arguments as well as assistant messages, including streaming responses. In GPT-6.2 Astra and Sol, protected reasoning is tied to the conversation history that produced it. If an earlier message is rewritten, the API drops the reasoning that followed it.

Two security specialists in business attire stand in a sunlit lobby as one hands a crimson binder to the other.
Illustration: AI & Tech News

OpenAI takes weekly action against account networks

OpenAI says it also investigates coordinated extraction attempts and takes action against networks of accounts when it identifies violations. Its weekly enforcement actions against such campaigns have ranged from thousands to tens of thousands of accounts this year. The company says it shares indicators of compromise with other model providers.

OpenAI is now applying selected safeguard improvements to GPT-6 Astra. It says it plans to share more about recent campaigns it has disrupted.

What this means

Following the November 17, 2026 release of GPT-6.2 Astra and Sol for ChatGPT Work, Codex, and API users, OpenAI detailed technical safeguards that tie protected reasoning to conversation history and drop reasoning when earlier messages are rewritten. The company noted particular concern regarding cybersecurity and biology capabilities, four days after announcing expanded cyber defense work with NATO, European agencies, and Ukraine using its tools to find and fix vulnerabilities. To address unauthorized extraction networks, OpenAI conducts weekly enforcement actions targeting thousands to tens of thousands of accounts while sharing indicators of compromise with other model providers.

Source: OpenAI

Newsletter

Get the best AI & tech news daily

A concise daily digest. Unsubscribe anytime.

We use your email only to send this newsletter.

Keep reading