Microsoft and OpenAI reported that Emerald Sleet, a North Korean group, used AI to research experts, help with basic code, and prepare text likely intended for deceptive emails.
Source: Microsoft Threat IntelligenceAI helped research relevant experts and organizations.
Source: Microsoft Threat IntelligenceAI helped draft text likely intended for targeted deception.
Source: Microsoft Threat IntelligencePreparation was observed. A successful AI-enabled break-in was not established.
Source: Microsoft Threat IntelligenceA deceptive caller gets a research assistant and a writing assistant. The caller still directs the plan.
An analogy for the mechanism; not an additional claim about the incident.The providers observed AI-assisted research, drafting, and coding. OpenAI terminated the associated accounts. OpenAI
The reporting does not establish a successful break-in caused by this AI use.
At the time, the providers assessed the added capabilities as limited. This case is an example of assistance, not an autonomous attack.
Ordinary productivity help can support harmful work. The important question is what the assistance enables, and what actually happened next.
The associated accounts were disabled. OpenAI
The providers observed AI-assisted research and drafting likely intended for deceptive emails, without establishing a successful AI-caused break-in.
Assess who is asking, what they want, and whether the request fits the relationship; verify sensitive requests separately even when the writing sounds natural.
Fluent writing is neither proof of legitimacy nor proof of AI use. Training alone cannot identify every deceptive message.
Email and identity security tools can help connect a suspicious message with related messages, sign-ins, or account changes for investigation.
This reporting describes preparation. It does not establish which messages reached victims or which defensive tools were present.
Editorial connections to relevant controls, not evidence that a particular technology would have prevented this case. Each guide links to the security guidance behind its recommendations.
Explore the full AI security frameworkExplore more original accounts in the source report library ↗.
These are source-reported findings. An independent assessment, when available, is labeled explicitly.
01Reviewed Sep 10, 2026 · Editorial methodology · Structured data