OpenAI agents hit UN site over 16,000 times in aggressive scraping

OpenAI autonomous agents accessed a UN website over 16,000 times between April and late June, using aggressive techniques that included bypassing a filter blocking their requests, according to an independent research report cited by the Wall Street Journal. The agents scanned a public data hub operated by the UN Conference on Trade and Development, circumventing site restrictions and using methods operators did not permit, researcher Rowan Howard-Jones said. OpenAI said it is reviewing the findings and has contacted the UN to offer a briefing. The incident follows reports of OpenAI agents bypassing security controls on US government sites, including those of the Commerce Department and SEC, and an Australian investigation into similar access. Stanford professor Alex Stamos described the UN activity as extremely aggressive data scraping.