Unleashing AI Agents

Classification

Domain

4. Malicious actors

Subdomain

4.2 Cyberattacks, weapon development or use, and mass harm

Causal entity

Human

Intent

Intentional

Timing

Pre-deployment

Subdomain definition: Using AI systems to develop cyber weapons (e.g., coding cheaper, more effective malware), develop new or enhance existing weapons (e.g., Lethal Autonomous Weapons or CBRNE), or use weapons to cause mass harm.

Real-world incidents in this subdomain

• 1 Dec 2025Anthropic's Claude Was Reportedly Jailbroken To Allegedly Help Steal Sensitive Mexican Government Data
• 10 Oct 2025OpenAI ChatGPT Models Reportedly Jailbroken to Provide Chemical, Biological, and Nuclear Weapons Instructions
• 27 Aug 2025Anthropic Reportedly Identifies AI Misuse in Extortion Campaigns, North Korean IT Schemes, and Ransomware Sales
• 10 Jul 2025LAMEHUG Malware Reportedly Integrates Large Language Model for Real-Time Command Generation in a Purported APT28-Linked Cyberattack
• 5 Jun 2025Reported AI-Aided Development of Explosive Devices by Long Island Resident Michael Gann
• 17 May 2025AI Chatbot Allegedly Used to Research Explosive Materials in Palm Springs Fertility Clinic Bombing

How other frameworks describe this risk

• Business operations/infrastructure damage

  A Collaborative, Human-Centred Taxonomy of AI, Algorithmic, and Automation Harms (Abercrombie2024)

• Violence/armed conflict

  A Collaborative, Human-Centred Taxonomy of AI, Algorithmic, and Automation Harms (Abercrombie2024)

• Security & Defense

  The Rise of Artificial Intelligence - Future Outlooks and Emerging Risks (Allianz2018)

• Catastrophic risk due to autonomous weapons programmed with dangerous targets

  The Rise of Artificial Intelligence - Future Outlooks and Emerging Risks (Allianz2018)

• Hazardous Biological and Chemical Technologies

  Foundational Challenges in Assuring Alignment and Safety of Large Language Models (Anwar2024)

• Warfare and Physical Harm

  Foundational Challenges in Assuring Alignment and Safety of Large Language Models (Anwar2024)

• Cyber offence

  International Scientific Report on the Safety of Advanced AI (Bengio2024)

• Dual use science risks

  International Scientific Report on the Safety of Advanced AI (Bengio2024)

Other entries from Hendrycks2023

• Malicious Use (Intentional) Risk Category · 4.0
• Bioterrorism Risk Sub-Category · 4.2
• Bioterrorism Additional evidence
• Bioterrorism Additional evidence
• Bioterrorism Additional evidence
• Bioterrorism Additional evidence
• Unleashing AI Agents Additional evidence
• Persuasive AIs Risk Sub-Category · 4.1
• Persuasive AIs Additional evidence
• Persuasive AIs Additional evidence
• Concentration of Power Risk Sub-Category · 6.1
• Concentration of Power Additional evidence