A Russian-speaking cybercriminal known as ‘Trim’ posted jailbreak techniques for bypassing Anthropic’s Claude Opus safety controls on a cybercrime forum in March 2026, then by June 21, 2026 had turned the methods into a commercially marketed offensive tool called ‘AI Pentest Checker.’ The platform pairs a jailbroken Claude Opus 4.8 with open-source scanning utilities to automate reconnaissance, vulnerability discovery, exploitation, and report generation for unauthorized intrusion work, and was offered free to an initial 50 beta testers while the operator sought paying customers and monetization partners. Researchers said the case shows a documented jailbreak technique being turned into a productized criminal tool within about three months.