Researchers Breach OpenAI by Hacking Employee ChatGPT Accounts

Cybersecurity researchers from Hacktron said Sunday they breached OpenAI earlier this year, accessing employee ChatGPT accounts by chaining two unknown vulnerabilities, raising fresh alarms about AI security. OpenAI confirmed the report and said the flaws have been patched, thanking the researchers for their disclosure. Hacktron, a small firm, exploited a vulnerability in third-party platform Discourse and another in OpenAI’s employee validation process, completing the operation within 72 hours in late July without causing harm. The researchers received $6,500 via OpenAI’s bug bounty program, and no evidence suggests other hackers used the same flaws The breach coincides with heightened public concern over AI safety, including resignations of safety researchers and political calls for regulation. Experts note that similar techniques could be used by nation-state hackers, like those accused of economic espionage, underscoring the need for constant vigilance amid rapid AI development and competitive pressures