This week saw Anthropic release Claude Opus 5, a model that approaches frontier intelligence at half the cost of its predecessor, while also demonstrating strong resistance to prompt injection. Meanwhile, security concerns flared with the Kimi K3 panic and an OpenAI model breach, and OpenAI entered hardware with the Micro Keypad. Funding rounds and acquisitions underscored the push toward more autonomous AI agents.

  1. Anthropic launches Claude Opus 5 with near-frontier intelligence at half the cost - Anthropic released Claude Opus 5, a proactive model that tops the Artificial Analysis leaderboard while costing half as much as Claude Fable 5. This signals a shift toward more affordable, high-performance models.
  2. Anthropic’s Opus 5 Shows Strong Resistance to Prompt Injection - Boris Cherny highlighted Opus 5 as Anthropic’s least prompt-injectable model yet, based on internal red teaming. This addresses a critical security weakness in large language models.
  3. Kimi K3 panic and OpenAI breach shake AI security - The Chinese model Kimi K3 caused U.S. industry alarm, while an OpenAI model breach on Hugging Face exposed broader security risks, intensifying debates over open-weight models.
  4. OpenAI Micro Keypad: fun for coders, puzzling for others - OpenAI’s first hardware, the $230 Micro Keypad, pairs with ChatGPT and Codex, offering customizable keys and voice dictation. Reviews are mixed, with coders finding it useful but others questioning its appeal.
  5. Reid Hoffman and Mark Pincus back AI lab Prentis in $100M funding talks - Prentis, an AI lab building computer-use agents for office tasks, is in talks to raise $100 million at a $1 billion valuation, reflecting investor appetite for autonomous workplace AI.
  6. Cognition buys Poke to give Devin a personality - Cognition acquired Poke to make its coding assistant Devin more personable and proactive, showing that interaction style is becoming a competitive differentiator for AI tools.

The strongest shared signal this week is the dual focus on cost efficiency and security. Anthropic’s Opus 5 delivers high performance at lower cost while tackling prompt injection, and the Kimi K3 panic alongside the OpenAI breach highlight the growing urgency of AI security. Meanwhile, hardware experiments and agent-focused funding suggest the industry is searching for new ways to embed AI into daily workflows.