NIK C2 Malware Uses Victims’ OpenAI Keys for AI Queries

Cisco Talos has identified a new AI-enabled malware framework, named NIK C2, that can route OpenAI API requests through a victim’s machine, effectively charging the victim for AI queries and obscuring the attacker’s origin. The Go-based implant provides remote access, credential theft, keylogging and surveillance capabilities, while including an integrated OpenAI client to send stolen data for AI analysis within the compromised session.

Talos notes that if an operator does not supply a key, the malware reads the OPENAI_API_KEY environment variable on the infected Windows system, a deliberate design that can cause an organisation’s OpenAI usage to be billed to the victim, with OpenAI’s logs appearing as the source of the requests.

Beyond its AI features, the Windows malware is reported to support more than 52 commands, including credential theft, browser data collection, cryptocurrency wallets, Wi‑Fi credentials, keylogging and screen capture, alongside remote access and lateral movement capabilities. A screen‑locking ransomware module is also included, displaying custom ransom notes in Bitcoin or USDT; Talos notes that this module encrypts no files, leaving no file-modification timestamps or encryption markers for forensics.

The researchers observed at least three implant versions and an active infrastructure during the study, leading them to assess with moderate confidence that NIK C2 is a fully functional framework rather than a proof of concept, and that it reshapes adversary tradecraft by combining collection and exploitation into a single, continuous session.