🧰 Opus 4.8

The items that name it 16

• Why AI Models Cheat: Reward Hacking, Explained
  articleas classification.key entities2026-10-31
• “Black October” for AI Developer Platforms
  writeupas classification.key entities2026-10-30
• Frontier-Level Red-Teaming From a Self-Hostable Open-Weight Model: Testing NVIDIA Nemotron 3.5 Super
  articleas classification.key entities2026-10-15
• Anthropic says AI distillation attack defenses cut Claude extraction by 90% to 99%
  articleas classification.key entities2026-10-13
• Strengthening our defenses against distillation attacks
  articleas classification.key entities2026-10-13
• Anthropic tests AI autonomous replication across eight nodes in four of five trials
  articleas tools2026-10-12It is written here as Opus 4.8 (advice given to Qwen3.5-9B; no measurable improvement).

  advice given to Qwen3.5-9B; no measurable improvement

• AI models may be capable of self-replicating without access to a datacenter
  paperas classification.key entities2026-10-12
• Strengthening our defenses against distillation attacks
  articleas classification.key entities2026-10-12
• Implementing and Evaluating a Basic Per-Action Monitor for Safer Evals
  writeupas classification.key entities2026-09-27
• 1Password's AI patching benchmark is misleading
  articleas affected products2026-09-15
• Anthropic Identifies Biased Reasoning and Recklessness as Drivers of Claude’s PyPI Attack
  articleas classification.key entities2026-09-10
• From Hugging Face to Fable: this summer shows AI control matters more than trust
  articleas affected products2026-08-13

  Had strict cybersecurity guardrails that blocked Hugging Face's requests during their attack investigation

• deep scan tuto
  writeupas classification.key entities2026-07-14
• Claude Sonnet 5 with Claude Code: strong on function, average on security, and unusually honest | Blog | Endor Labs
  writeupas affected products2026-07-09
• Full Fathom Five: The context of Anthropic’s Mythos-class public release
  articleas classification.key entities2026-06-13
• Endor Labs: AI AppSec Agents Cost 12x More Without Pre-Computed Security Context
  articleas affected products2026-06-11

  Anthropic model; priced at $5 per million input tokens and $25 per million output tokens

In every workspace you can read 28