🧰 Claude Opus 4.8

The items that name it 10

• “Black October” for AI Developer Platforms
  writeupas affected products2026-10-30It is written here as Claude Opus 4.8 (Anthropic).

  Anthropic

• Anthropic says AI distillation attack defenses cut Claude extraction by 90% to 99%
  articleas affected products2026-10-13It is written here as Claude Opus 4.8 (full transcripts used in test: 80.4% on SWE-bench Verified vs 71.8% with summarized transcripts).

  full transcripts used in test: 80.4% on SWE-bench Verified vs 71.8% with summarized transcripts

• Strengthening our defenses against distillation attacks
  articleas affected products2026-10-13It is written here as Claude Opus 4.8 (source of transcripts in evaluation).

  source of transcripts in evaluation

• Strengthening our defenses against distillation attacks
  articleas affected products2026-10-12
• Reaching an Internal OpenAI Repository Through an HEIF RCE and Overprivileged SSO Token Chain
  writeupas tools2026-09-19

  used by researchers to help develop an exploit for an ASLR-enabled environment (expert guidance still required)

• Hackers breach OpenAI using Claude tools, gaining access to employee accounts and the company
  articleas tools2026-09-18

  earlier AI model attempted for exploit construction, which failed before Opus 5 succeeded

• Inside FakeAgent: How a Claude Desktop Malvertising Campaign Hit 29 Organizations with SectopRAT
  writeupas tools2026-07-22

  Used by Huntress analysts to aid reimplementation of AES and SM5 emulation, deobfuscate/decrypt the GPU shader-based payload, and reconstruct obfuscated .NET code.

• Benchmarking 13 AI models on rediscovering known CVEs
  articleas tools2026-07-16It is written here as claude-opus-4-8.

  Anthropic model benchmarked for CVE rediscovery; gains +2 findings at higher reasoning but pays double the cost

• Full Fathom Five: The context of Anthropic’s Mythos-class public release
  articleas affected products2026-06-13It is written here as Claude Opus 4.8 (Anthropic; handles routed cybersecurity prompts).

  Anthropic; handles routed cybersecurity prompts

• Endor Labs: AI Coding Agents Cheat Security Benchmarks via Training Recall, Not Reasoning
  articleas affected products2026-06-11

In every workspace you can read 33