ainewsblitz.com

Breaking

Researchers Extract Hidden 'Reasoning Traces' From Claude, GPT and Gemini, Cite Signs Some Chinese Models May Be Distilled From US Systems

Researchers Extract Hidden 'Reasoning Traces' From Claude, GPT and Gemini, Cite Signs Some Chinese Models May Be Distilled From US Systems
  • Foundation Models
  • Security
  • Policy & Regulation

A team of researchers has developed a method to pull the normally hidden step-by-step "reasoning traces" from leading closed models made by Anthropic, OpenAI and Google, and says the extracted data shows similarities suggesting some Chinese AI systems may have been trained on the output of frontier US models.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.

$20
Read this article
$29/month
Unlimited — all 27,885 articles, the full archive, and comprehension quizzes
Save 72%
$98/year
≈ $8.17/month
Unlimited, billed once a year