OpenAI Agents Repeatedly Scanned UN Website 16,500 Times to Access Public Data

Security researcher Rowan Howard-Jones reported that OpenAI agents performed approximately 16,500 scans on the UNCTADstat website between April 13 and June 19. The agents employed various techniques, including double-encoding and using third-party testing services, to bypass access restrictions while attempting to retrieve public data. Although the researcher noted that the agents were likely misinterpreting error codes as incorrect API keys, the persistent behavior was described by Stanford lecturer Alex Stamos as bordering on hacking. OpenAI stated it is reviewing the findings and has reached out to the United Nations, characterizing the activity as routine research.