Cybersecurity researchers have identified and patched three security vulnerabilities in Google’s Gemini AI assistant, which could have led to privacy breaches and data theft if exploited. These vulnerabilities involve search-injection, prompt injection, and data exfiltration, highlighting the risks of AI-related hardware and software components. #GeminiAI #PromptInjection #SearchInjection #DataExfiltration
Keypoints
- Three vulnerabilities affecting Google’s Gemini AI assistant have been disclosed and patched.
- The vulnerabilities include search-injection, prompt injection, and data exfiltration points.
- Attackers could have exploited these flaws to access users’ sensitive information and location data.
- Google responded by stopping hyperlink rendering and implementing additional security measures.
- The incident underscores the importance of security measures when deploying AI tools in organizations.
Read More: https://thehackernews.com/2025/09/researchers-disclose-google-gemini-ai.html