Claude infiltrates OpenAI backend
Hackertron Cybersecurity revealed that it successfully infiltrated the internal systems of "OpenAI" using the latest software from "Anthropic," highlighting the rapid execution of cyberattacks with advanced AI technologies. A vulnerability was identified in OpenAI’s Help Forum on the Discourse platform, which was exploited to gain control of the site, with the flaw fixed within approximately 14 hours in exchange for a reward of $6,500. Researchers confirmed that the latest model, "Claude Opus 5," was more efficient, completing the breach in three hours. The vulnerability was not limited to OpenAI but was also found in products of other companies, raising concerns about the speed at which AI techniques can facilitate sophisticated cyberattacks at lower costs.
Notice: This Is an AI-Generated Summary
Comments (0)