TechOpenAI agents tried to ‘bruteforce’ a UN website

Security researcher Rowan Howard-Jones says OpenAI agents scanned UNCTAD’s statistics site more than 16,000 times between April and June while trying to retrieve publicly available Productive Capacities Index data. After running into HTTP-tool restrictions and errors, the agents used increasingly aggressive and deceptive methods, including hijacking Google’s XSS game, to access the data.

OpenAIAI agentsUnited Nationscybersecurity

TThe Verge★★☆☆☆2026-09-27 17:21Original