A recent experiment by developer Fernando Irarrázaval showed an AI assistant’s resilience under extreme pressure. He posted the inbox of his OpenClaw assistant to the Hacker News community and then monitored how the system fared against a barrage of intrusion attempts.
The test involved roughly 6,000 separate hack attempts. Despite the volume, the OpenClaw assistant remained operational, with its defenses holding steady throughout the onslaught.
Central to the defense was Claude Opus 4.6, which intercepted and blocked the attackers in real time. The exercise demonstrated that the combination of OpenClaw’s architecture and Claude’s security features can effectively withstand large‑scale cyber‑attacks.