Google is bringing generative artificial intelligence to more practical applications, moving away from the realm of image and video generation to focus on cybersecurity. In this sense, the company has introduced Google Threat Intelligence, an innovative solution that combines the expertise of cybersecurity company Mandiant and the VirusTotal database with the powerful Gemini model.
According to Google, in a blog post, this new product will be based on its Gemini 1.5 Pro language model, which has shown a remarkable ability to reduce the time required for malware analysis. For example, the AI took only 34 seconds to analyze the code of the WannaCry ransomware, used in several cyber attacks in 2017.
One of the most outstanding features of Google Threat Intelligence is its ability to summarize threat reports in natural language, allowing companies to understand the potential impact of attacks more clearly and quickly, and take appropriate response measures. In addition, it will provide a wide network of information to monitor potential threats and prioritize necessary actions.
Google has integrated Mandiant into the project to assess security vulnerabilities in AI projects, which is in line with the company’s Secure AI Framework. This framework aims to ensure the security of AI models against potential malicious attacks, such as data poisoning, thus reinforcing the protection of digital infrastructures.
However, Google is not alone in this initiative, as Microsoft has also taken its first steps in the field of cybersecurity with Copilot for Security, a chatbot that combines the GPT-4 models from OpenAI and Microsoft’s Copilot and allows cybersecurity professionals to ask direct questions about threats.
