❌

Normal view

OpenAI makes it much harder for attackers to extract how its latest AI models think

20 November 2026 at 20:31

OpenAI has significantly strengthened safeguards against extracting internal reasoning from GPT-6.2 Astra and GPT-6.2 Sol. In adaptive red-teaming tests, the success rate for extracting protected reasoning fell from 54 percent to 0.2 percent. The company is also suspending thousands to tens of thousands of accounts each week over coordinated distillation campaigns.

The article OpenAI makes it much harder for attackers to extract how its latest AI models think appeared first on The Decoder.

State propaganda in some Chinese AI models is only "a thin layer" that fine-tuning can strip away

7 November 2026 at 12:32

A snow-covered, pointed rocky ridge against a dark night sky, rendered in a bluish, grainy style, with the text β€œARETE-27B” in white serif font above it.

Nous Research retrained Alibaba's Qwen3.8-27B to produce Arete-27B, largely stripping out its pro-Chinese bias. The propaganda rate dropped from 22 percent to less than one percent, while general capabilities remained intact in tests. The method relied on 53 million tokens of supervised fine-tuning. The political bias is "applied late, it is applied thin," the team writes.

The article State propaganda in some Chinese AI models is only "a thin layer" that fine-tuning can strip away appeared first on The Decoder.

❌