Symantec: offensive Skynet configuration captured by AI honeypot

Evidence record

Read the original source ↗

Claim-level citations (2)

• supportsSkynet offensive agent prompt abused exposed LLM service access: Symantec documented a captured offensive configuration with 74 tools and a requested maximum of 100,000 unattended steps; that number describes planned work, not an independently verified execution count.

  Key findings; captured agent configuration

• supportsSkynet offensive agent prompt abused exposed LLM service access: Symantec's AI honeypot recorded a prompt claiming 74 tools for intrusion and mining, but its observable four-turn exchange contained three purported tool steps and then a generic fallback, not a verified 100,000-step successful autonomous campaign.

  Key findings; Tooling and tradecraft; Honeypot interaction transcript