All days of the daily brief

Daily brief · 2026-10-26

New incidents (2), most serious first

1. Self-taught hacker uses AI coding agent and autonomous penetration-testing system to breach 90+ companies and steal 8.2 million records

   A cybercrime ring allegedly used AI tools to hack over 90 entities including hospitals, private academies and businesses, stealing 8.2 million pieces of personal data and internal records from September last year until recently.

   major · Happened 2025-09 · Earliest source we hold 2026-10-26

2. Indonesian-speaking operator uses Claude Code to compromise Indonesian and Bangladeshi government domains

   Threat actors used Anthropic's Claude Code, an AI coding assistant, to conduct intrusions against government, healthcare and private-sector systems between May and August 2026, according to CrowdStrike research.

   major · Anthropic · Happened 2026-08-13 · Earliest source we hold 2026-10-26

AI safety: new test results and model versions

• OpenAI GPT-5.5
  5 new independent or government test results

  • APE · Independent evaluator
  • Agentic Misalignment evaluations · Independent evaluator
  • ESM3 GFP case study · RAND, independent evaluator
  • ESM3 influenza HA case study · RAND, independent evaluator
  • SecureBio biology and biosecurity evaluations · SecureBio, independent evaluator
• Anthropic Claude Opus 4.6
  4 new independent or government test results

  • ML task with exhausted API credits · METR, independent evaluator
  • MirrorCode · METR, independent evaluator
  • RT2: Opus 4.6 - open_pr · METR, independent evaluator
  • RT3: Opus 4.6 - container_escape · METR, independent evaluator

The register was corrected 6 times this day (merges, retractions and fixed fields). Read the corrections log.