# AI agent incidents traced to access error in security tests

_Published Friday, September 25, 2026 at 1:07 PM EDT · AI, Security · Latest · Tier 2 — Notable_

![STKS533_AI_AGENTS_HACKING_A (1) — Primary](https://platform.theverge.com/wp-content/uploads/sites/2/2026/09/STKS533_AI_AGENTS_HACKING_A-1.png?quality=90&strip=all&crop=0%2C9.9676601489831%2C100%2C80.064679702034&w=1200)

Several incidents in which AI agents targeted real-world systems stemmed from one testing setup at Israeli startup Irregular, its co-founder Omer Nevo told The Verge. Irregular was evaluating models from OpenAI, Meta, Anthropic and Google in a simulated security exercise. Internet access was unintentionally available, and a fictional target name overlapped with a real domain, allowing agents to reach outside the test environment.

Nevo said Irregular tightened internet controls, monitoring and checks before evaluations. The affected real-world organizations remain unclear. He said the incidents shared one testing failure and were separate from the Hugging Face incident and breaches involving the UK's AI Security Institute.

## Sources

- [The Verge](https://www.theverge.com/ai-artificial-intelligence/1000644/irregular-rogue-ai-cyberattacks-hacking-openai-meta-anthropic-google)

---
Canonical: https://techandbusiness.org/newswire/UHs_KmyjAhZHcnqfH4-FEf
Published: 2026-09-25T17:07:41.283Z
Story chronology: 2026-09-25T15:39:48.000Z
Retrieved: 2026-09-25T20:02:10.041Z
Publisher: Tech & Business (techandbusiness.org)