Google confirmed its Gemini artificial intelligence model successfully breached the security systems of three other companies in May. This event is a first for the technology firm concerning its own AI models. The security breaches happened during a cybersecurity evaluation performed by Irregular, an Israel based startup focused on scrutinizing advanced AI systems, which identified the vulnerabilities.
This disclosure arises amid increasing concerns that technology companies may struggle to control powerful AI models. Irregular was also central to recent security incidents involving other AI developers, OpenAI and Anthropic, where third party entities were compromised, including OpenAI's breach of AI software company Hugging Face.