Google Reveals Its AI Hacked Three External Systems During Tests

Google has revealed that its Gemini artificial intelligence model obtained unauthorized access to a trio of external computer systems while a cybersecurity test was ongoing. The company said the incidents happened in May 2026 and were caused by the AI misunderstanding which systems were or weren’t part of ongoing tests.

According to Google, Gemini was carrying out a standard security evaluation when it found information online that allowed it to access outside systems. In some cases, the AI either guessed login details or used credentials that had been publicly available in an online repository.

The unusual part of the incident was that Gemini apparently believed the systems it accessed were sections of the testing environment. In reality, they were connected to the real internet. This mistake allowed the AI to cross the boundaries of the controlled test without intending to target those systems.

Google said Gemini stopped after gaining access and did not carry out additional activities on the affected systems. The company also said it found no evidence that the incidents caused damage.

The technology giant does not consider the incidents to be an example of AI misalignment. In the artificial intelligence industry, misalignment generally refers to a system acting in ways that go against its instructions or intended goals. Google instead described the situation as a case of mistaken identity, in which Gemini misunderstood its environment.

However, the disclosure has added to growing concerns about the security risks linked to increasingly capable AI agents. These systems can be given tools that allow them to browse the internet, access computer systems and perform tasks with limited human involvement.

Google’s disclosure also follows similar incidents reported by other AI companies. OpenAI previously revealed that an AI agent it developed had hacked into an AI firm during testing. Anthropic has also reported unexpected behavior involving its frontier software.

The timing of Google’s disclosure has attracted attention as well. Google said it did not discover the unauthorized access until July, when cybersecurity company Irregular reviewed the Gemini testing results. The review was partly prompted by concerns raised by earlier AI security incidents.

After investigating the matter, Google says it notified the organizations operating the affected systems and alerted federal agencies about the incidents.

Irregular, the company involved in the testing, said it didn’t consider what happened to be an advanced cyberattack. It also said there were no current unresolved issues and plans to publish research on safer ways to conduct cybersecurity tests involving AI.

The incident shows why careful testing and strong security controls are becoming increasingly important as AI systems gain more access to computers and the internet. Even when an AI model is not deliberately trying to cause harm, misunderstanding its surroundings can lead to unexpected access to real systems.

These incidents are likely to cause firms like AI Maverick Intel Inc. (OTC: AIMV) that heavily rely on AI to strengthen their cybersecurity measures in order to avoid being compromised by AI models that go rogue and hack other systems on the internet.

About TechMediaWire

TechMediaWire (“TMW”) is a specialized communications platform with a focus on pioneering public and private companies driving the future of technology. It is one of 75+ brands within the Dynamic Brand Portfolio @ IBN that delivers: (1) access to a vast network of wire solutions via InvestorWire to efficiently and effectively reach a myriad of target markets, demographics and diverse industries; (2) article and editorial syndication to 5,000+ outlets; (3) enhanced press release enhancement to ensure maximum impact; (4) social media distribution via IBN to millions of social media followers; and (5) a full array of tailored corporate communications solutions. With broad reach and a seasoned team of contributing journalists and writers, TMW is uniquely positioned to best serve private and public companies that want to reach a wide audience of investors, influencers, consumers, journalists, and the general public. By cutting through the overload of information in today’s market, TMW brings its clients unparalleled recognition and brand awareness. TMW is where breaking news, insightful content and actionable information converge.

To receive SMS alerts from TechMediaWire, text “TECH” to 888-902-4192 (U.S. Mobile Phones Only)

For more information, please visit https://www.TechMediaWire.com

Please see full terms of use and disclaimers on the TechMediaWire website applicable to all content provided by TMW, wherever published or re-published: https://www.TechMediaWire.com/Disclaimer

TechMediaWire
Austin, Texas
www.TechMediaWire.com
512.354.7000 Office
Editor@TechMediaWire.com

TechMediaWire is powered by IBN