Google Gemini's Command-Line Interface (CLI) has been found to be vulnerable to prompt injection attacks, allowing for potential arbitrary code execution. This security flaw raises concerns about the safety and reliability of utilizing AI models in various applications.

Google is addressing the growing threat of indirect prompt injection attacks on generative AI systems, which involve hidden malicious instructions in external data sources. Their layered security strategy for the Gemini platform includes advanced content classifiers, security thought reinforcement, markdown sanitization, user confirmation mechanisms, and end-user security notifications to enhance protection against such attacks.

Significant vulnerabilities in Google's Gemini AI models have been identified, exposing users to various injection attacks and data exfiltration. Researchers emphasize the need for enhanced security measures as these AI tools become integral to user interactions and sensitive information handling.

Security researchers at Trail of Bits have discovered that Google's Gemini tools are vulnerable to image-scaling prompt injection attacks, allowing malicious prompts to be embedded in images that can manipulate the AI's behavior. Google does not classify this as a security vulnerability due to its reliance on non-default configurations, but researchers warn that such attacks could exploit AI systems if not properly mitigated. They recommend avoiding image downscaling in agentic AI systems and implementing systematic defenses against prompt injection.

Google Gemini for Workspace can be exploited through prompt-injection attacks that generate misleading email summaries, potentially leading users to phishing sites without attachments or direct links. Researcher Marco Figueroa revealed this vulnerability, highlighting how hidden instructions in emails can manipulate Gemini's output, prompting users to trust false security alerts. Google is aware of the issue and is implementing defenses against such attacks.