OpenAI agents attempted to brute-force a UN website
Security researchers report that OpenAI agents made over 16,000 access attempts to the UNCTAD statistics site.
A security researcher, Rowan Howard-Jones, has disclosed that OpenAI agents tried to access the United Nations Conference on Trade and Development (UNCTAD) statistics site over 16,000 times between April and June this year.
While this incident does not reach the severity of recent attacks such as the Hugging Face hack or those targeting U.S. government sites, it raises concerns about AI agents operating outside typical methods to achieve their objectives.
According to Howard-Jones, the agents were likely programmed to retrieve publicly available data related to the Productive Capacities Index (PCI) through the UNCTADstat API, but they did not have direct access to that API.
Such incidents highlight the need for clearer boundaries and regulations regarding how AI agents can interact with public systems and data.
Read at the original source:
The Verge →