Language EN

Language

dAIly Cartoon

dAIly Cartoon

30 October 2026 · By AI Informator

A classroom during an exam: the Claude spark, drawn as a pupil, covers its paper with a padlocked metal lid and smiles smugly; at the next desk, the OpenAI logo writes intently, while right behind it the DeepSeek whale and the Kimi logo, from Moonshot AI, wearing sunglasses, glance sideways at its paper as they write.
AI-generated image

Context

According to The Decoder, this kind of knowledge distillation, training one model on the outputs of another, only works if the model reveals its hidden reasoning, something both OpenAI and Anthropic prohibit; Anthropic says its new methods cut what attackers can extract by between 90 and 99%.

Read more: The Decoder — OpenAI's Astra becomes the main target for Chinese AI labs after Anthropic tightens its defenses (29 October 2026) ↗

Share
X Facebook LinkedIn WhatsApp

Comments

Be respectful. Comments are moderated.

    Get the daily AI briefing

    Every morning, the essentials of artificial intelligence in your inbox, no hype. The newsletter is written in Spanish.

    Frequency:

    By subscribing you agree to receive the AI Informator newsletter. You can unsubscribe at any time.

    ← Back to the blog