Ledger / Jan 2025
APT43 uses public AI chatbots and image generators for research and lure material
Strong
GTIG detected APT43 accessing public LLM and image tools and infers research use; it says the intended purpose is not clear.
- Model
- Not namedAI chatbot (model not named), AI image generator (model not named)
- Agent harness
- Nonechat or tool use only
- Attributed country
- North Koreaas attributed by Google Threat Intelligence Group
- Sectors hit
- government, defenseNot stated in the source
Split from GTIG's January 2025 report on nine North Korean groups: GTIG detected APT43 accessing several public AI chatbots, including for research on South Korean military and nuclear affairs, and AI image tools that make realistic human portraits.
“GTIG has identified APT43 actors accessing multiple publicly available AI image generation tools, including tools used for image manipulation and creating realistic-looking human portraits.”Google Threat Intelligence Group
Sources
- Adversarial misuse of generative AIGoogle Threat Intelligence Group · 29 Jan 2025
- Date
- Jan 2025
- Actor
- APT43 (nation state), as attributed by Google Threat Intelligence Group
- Nexus
- North Korea
- Targets
- Not stated in the source
- What the AI did
- Recon, Social engineering
- Model as named
- AI chatbot (model not named), AI image generator (model not named)
- Model basis
- A researcher names it
- Autonomy
- Assisted a human
- Outcome
- GTIG saw the tools accessed; it says the intended purpose is not clear.