Redacted Risk Report Feb 2026
www-cdn.anthropic.com · 6,994 words · saved by 1 readers
N/A
Risk Report: February 2026 anthropic.com 1 Introduction and executive summary 5 1.1 Structure of the report 6 1.2 Executive summary of findings 7 2 Autonomy threat model 1: Sabotage 12 2.1 Overview 12 2.2 Threat model…
related reading
- Countering misuse of AI: September 2026 / Anthropicanthropic.com
- Claude Mythos Preview System Cardwww-cdn.anthropic.com
- Frontier Risk Report (February to March 2026) - METRmetr.org
- metr.org/risk-report-feb-mar-2026.pdf#page=33.47metr.org
- Anthropic February 2026 Risk Report: SecureBio's External Reviewsecurebio.org
- Teaching Claude Whyalignment.anthropic.com
- Agentic misalignment: How LLMs could be insider threats \ Anthropicanthropic.com
- Agentic Misalignment in Summer 2026alignment.anthropic.com
- An alignment assessment of recent cybersecurity incidentsanthropic.com
- Teaching Claude why \ Anthropicanthropic.com
- Recommendations for Technical AI Safety Research Directionsalignment.anthropic.com
- Natural-emergent-misalignment-from-reward-hacking-paper.pdfassets.anthropic.com