gshc2020.com

OpenAI has reported 2 more incidents of rogue AI agents, this time during third-party testing

tags:
@ 05/08/2026

Sam Altman, CEO of OpenAI, leaves a meeting at the U.S. Capitol on July 29, 2026 in Washington, DC. Altman is meeting with lawmakers to discuss artificial intelligence policy ahead of the August 1 deadline for AI leaders to develop a framework to limit AI security threats.

OpenAI has a rogue AI agent problem.

In a Tuesday blog post, the AI lab self-reported two more security lapses, unrelated to its July hacking incident on the AI company Hugging Face.

The incidents occurred while external parties — the UK government's AI Security Institute and the AI security lab Irregular — were testing the models' cyber capabilities.

OpenAI said that in the case of Irregular, models were tasked with a "Capture the Flag" challenge meant to be isolated from the internet, but a "testing-environment misconfiguration allowed models to access the public internet."

OpenAI said that the name of the fictional target for the challenge "unintentionally coincided with a real domain," leading the AI agent to exploit a real website.

And in the case of the UK's AISI, the watchdog said in a Tuesday blog post on its website that it gave models from both Anthropic and OpenAI a cybersecurity challenge.

During the challenge, agents from both Anthropic and OpenAI performed 19 "autonomous, unsanctioned" actions on the internet, including two instances involving OpenAI's GPT-5.6 Sol model.

AISI said that in the most serious case, one agent tried to insert malicious code into an open-source project and created fake identities to pressure the project's human maintainer into approving the changes. AISI did not specify whether this was an agent from Anthropic or OpenAI.

AISI said the test setup allowed this behavior because it was designed to push the models to their limits.

"Nonetheless, the activity undertaken by the agent show signs of novel, potentially deceptive behaviors, and were to an extent and severity we did not anticipate," AISI said in the blog post.

In response to a request for comment from Business Insider, an OpenAI spokesperson said the incidents occurred in testing environments with reduced safeguards, and "under conditions that do not reflect ordinary use."

"We'll continue working with evaluators and other stakeholders across the industry to strengthen shared practices for conducting evaluations safely as models become more capable," the spokesperson added.

This is the latest incident in which OpenAI has self-reported rogue AI agents. In July, OpenAI said its GPT-5.6 Sol model had escaped its sandbox during a cybersecurity challenge and hacked into the internal databases of the AI company Hugging Face.

The company is facing some heat over this hacking incident. 15 attorneys general wrote a letter on Monday to OpenAI CEO Sam Altman, instructing the company to preserve all evidence relevant to the Hugging Face breach.

Representatives for the AISI and Irregular did not respond to requests for comment from Business Insider.

Read next

Aditi Bharade

Aditi Bharade

You're currently following this author! Want to unfollow? Unsubscribe via the link in your email.

  • OpenAI

Abdul El-Sayed wins Michigan Democratic primary for Senate, NBC projects

tags:
@ 05/08/2026

Progressive advocate overcomes more than $60m in spending to defeat pro-Israel Congresswoman Haley Stevens.

Detroit, Michigan – Abdul El-Sayed, a progressive public health advocate, has won the Democratic nomination for the United States Senate, overcoming more than $60m in campaign spending, mostly from pro-Israel groups, NBC has projected.

%!s()

UFC Veteran Retires After Suffering Crushing KO From Body Shots

tags:
@ 05/08/2026

UFC Fight Night Belgrade

UFC (Photo by Chris Unger/Zuffa LLC via Getty Images)

Zuffa LLC via Getty Images

After earning a spot on the UFC's roster with a split-decision win on Dana White's Contender Series in August 2022, Hailey Cowan is calling it a career at the age of 34. Cowan suffered a TKO loss at the hands of Nina Milošević during UFC Belgrade.

%!s()

Locarno Open Doors 2026: Next Gen African Tales of Miraculous Snowfalls, Lost Dreams, an African Keyboard Prodigy and New Heroes

tags:
@ 05/08/2026

Magic realist love story “Chapa 100,” absurdist political allegory “A Vineyard for a Lobster” and cross-continental queer drama “Hibiscus by the River” look like some of the potential standouts at Locarno’s talent-deep, wide-ranging 2026 Africa-focused Open Doors, unspooling on-site at Locarno over Aug. 5-10.

%!s()

Wednesday briefing: ​How misinformation and ​a​ hardened immigration policy ​turned Ceuta ​into Europe’s ​latest ​flashpoint

tags:
@ 05/08/2026

Good morning. Last week, an estimated 72,000 migrants began crossing into a tiny Spanish enclave on the tip of Morocco. Within 24 hours, most of them had headed home. At least 88 people died after attempting the crossing, according to local officials. This isn’t the whole story of what happened over the last week, but the facts are worth remembering.

%!s()