AI & Tech Developments - Feb 24

  • 18:56 — Anthropic catches Chinese labs stealing Claude at scale, marking the largest AI heist in history. @ns123abc
  • 22:31 — Discussion on how AI assistants like Claude can seem human through persona selection. @AnthropicAI
  • 02:20 — Introduction of a new API feature that enhances agent performance. @OpenAIDevs
  • 02:59 — Instructions on how to use Claude to orchestrate Codex agents effectively. @om_patel5
  • 03:08 — Codex compaction endpoint is literally voodoo, improving agent performance. @LLMJunky

📱 Source Tweets

We’ve identified industrial-scale distillation attacks on our models by DeepSeek, Moonshot AI, and MiniMax. These labs created over 24,000 fraudulent accounts and generated over 16 million exchanges with Claude, extracting its capabilities to train and improve their own models.

@AnthropicAI

AI assistants like Claude can seem shockingly human—expressing joy or distress, and using anthropomorphic language to describe themselves. Why? In a new post we describe a theory that explains why AIs act like humans: the persona selection model.

@AnthropicAI

Introducing Nova. A futuristic canvas for your personal development. Download on macOS, Windows and Linux.

@try_nova_app

HOW TO LET CLAUDE CODE SPIN UP CODEX AGENTS most people think you have to choose one ai coder you don’t you can make claude orchestrate codex like a specialized coding agent here’s how to actually do it 1\ install codex cli first install codex locally npm i -g

@om_patel5

Introducing WebSockets in the Responses API. Built for low-latency, long-running agents with heavy tool calls. http://developers.openai.com/api/docs/guides/websocket-mode…

@OpenAIDevs

BREAKING: Anthropic Catches Chinese Labs Stealing Claude at Scale - Largest AI Heist in History! Three Chinese labs ran coordinated distillation attacks. Anthropic traced accounts directly to researchers at each lab and dropped receipts: >DeepSeek: 150k exchanges, used claude

@ns123abc

Discussion on how AI assistants like Claude can seem human through persona selection. @AnthropicAI

@AnthropicAI

Introduction of a new API feature that enhances agent performance. @OpenAIDevs

@OpenAIDevs

Instructions on how to use Claude to orchestrate Codex agents effectively. @om_patel5

@om_patel5

Codex compaction endpoint is literally voodoo. I used to reset my context window at 50% religiously because that's what I had to do with other agents. Until one day I heard @steipete say "no, what are you doing that for?" It was a hard habit to break at first because I had

@LLMJunky