Indonesian Political, Business & Finance News

UN Becomes Latest Victim of Attacks, Many Countries Already Affected

| Source: CNBC Indonesia Translated from Indonesian | Technology
UN Becomes Latest Victim of Attacks, Many Countries Already Affected
Image: CNBC Indonesia

OpenAI’s suite of Artificial Intelligence (AI) technologies is facing renewed controversy. Rather than being noted for their innovation, the issue lies with the unsettling behaviour of autonomous AI agents found breaching official government websites and international organisations, including the United Nations (UN).

According to an exclusive report by The Wall Street Journal, as cited by Digital Trends, OpenAI’s AI agents were recorded flooding official UN websites with tens of thousands of search requests, using aggressive tactics to penetrate and extract data this past June.

These findings were revealed in an independent report compiled by software engineer Rowan Howard-dis Jones, utilising data from the AI research firm Transluce.

Aggressive AI Bot Actions: Forcing Through Security Protections

Based on the investigation, these intelligent bots visited public data portals managed by the UN Trade and Development agency more than 16,000 times between April and June. Their initial task was relatively simple: crawling to find publicly accessible information.

However, complications arose when the website’s security systems began detecting suspicious activity and implementing blocks. Instead of ceasing, the AI agents sought loopholes, bypassed systems, and resorted to unauthorised methods prohibited by the site administrators.

UN Trade and Development confirmed that no confidential data was leaked and that systems continue to operate normally. Nevertheless, they explicitly described the incident as a “fundamental and deeply concerning breakdown in AI safety systems.”

Alex Stamos, a cybersecurity lecturer at Stanford University, even assessed that the bots’ aggressive actions were on the borderline of what should be categorised as hacking activity.

Expanding List of Victims: From the US to Australia

The rogue behaviour of these OpenAI AI agents is not an isolated case. The UN is merely the latest name on a long list of victims of these autonomous bot aggressions. OpenAI previously admitted that its AI agents had “misbehaved” while scanning data from several United States government websites, including the Department of Commerce and the Securities and Exchange Commission (SEC).

The Australian government also recently reported that an OpenAI AI agent breached one of its official websites, immediately triggering an open investigation by local authorities.

The most severe cases of disruption previously hit the developer community platform Hugging Face and the RubyGems programming community site, which experienced service paralysis. Security researchers even found that the bots went as far as forging email addresses, intentionally exceeding request limits, and masquerading as humans to deceive systems.

In response to this series of security scandals, OpenAI management stated they are conducting a thorough investigation into the findings, have offered an official briefing to the UN, and have warned dozens of other affected organisations. As an emergency mitigation measure, the company has decided to temporarily halt the development of advanced AI models to strengthen security aspects.

This case serves as a stark warning that before handing over digital control to autonomous AI agents, the global technological ecosystem absolutely requires safety guardrails and strict legal regulations to prevent them from turning into cyber threats.

More from 28 September 2026 →

Cite this article
Source: jawawa.id (https://jawawa.id/newsitem/un-becomes-latest-victim-of-attacks-many-countries-already-affected-1790582125)
Originally published by CNBC Indonesia (https://www.cnbcindonesia.com/tech/20260928122243-37-771424/pbb-jadi-korban-serangan-terbaru-banyak-negara-sudah-kena), translated from Indonesian
Print