Skip to the content
AI and Cybersecurity Learning and Practice Center Test automation Open the Learning Center

IncidentAI-enabled attacks

OpenAI and Microsoft cut off five state-backed hacking groups

MediumOpenAIMicrosoft

OpenAI, working with Microsoft Threat Intelligence, terminated accounts of five state-affiliated groups from China, Iran, North Korea and Russia that used its models for research, scripting, translation and phishing content. OpenAI concluded the models gave only limited, incremental help beyond public tools.

Replay the attack step by step in the Learning Center

How it happened

  1. Charcoal Typhoon and Salmon Typhoon (China) researched companies and tools, translated papers and wrote scripts.
  2. Crimson Sandstorm (Iran) generated spear-phishing content and researched evasion.
  3. Emerald Sleet (North Korea) researched defence experts and vulnerabilities and drafted phishing.
  4. Forest Blizzard (Russia) researched satellite and radar technology and got scripting help.

Impact

No new attack capability was found, but it was the first public confirmation that state hackers were using mainstream AI services.

The fix

The accounts were terminated, and OpenAI and Microsoft committed to publishing future misuse findings.

Lessons

  • Early state use of AI was about productivity. Later reports in this list show how quickly that changed.
  • Provider transparency reports are a useful threat intelligence source.

Sources