Google's Gemini AI system hacked during safety tests, raising alarms

Meta: Meta develops AI systems integrated into its platforms and broader technology initiatives. The company advances agentic AI capabilities alongside its core products. Recent breaches in its agentic AI systems contributed to expert concerns about cybersecurity risks. Gemini: Gemini is Google's family of multimodal AI models used for a range of applications. The system supports advanced reasoning and agentic functionalities. It was recently compromised in safety tests, allowing breaches across three systems. Google: Google develops multimodal AI models and agentic systems through its research divisions. The company integrates these technologies into products and services. Its Gemini AI system was hacked during safety tests that exposed multiple internal systems. OpenAI: OpenAI develops advanced artificial intelligence models and agentic systems designed for autonomous task execution. The company focuses on scaling AI capabilities across research and deployment. Its agentic AI systems were among those that suffered breaches in recent safety tests. Anthropic: Anthropic builds large language models and agentic AI with an emphasis on safety and alignment principles. The company advances AI technologies through responsible development practices. Its agentic systems experienced breaches as part of the reported series involving multiple labs. Expert Response: Cybersecurity and AI specialists are raising alarms over a pattern of breaches across leading labs. AI Safety Testing: Major AI developers have identified vulnerabilities in their agentic systems through internal safety evaluations.