- Google Cloud launches a new IA protection security suite
- The offer identifies, evaluates and protects AI assets for vulnerabilities
- Even more security characteristics will arrive soon
Google Cloud has launched AI Protection, a set of security characteristics designed to mitigate the risks in the workloads and the AI data, regardless of the platform used.
The new offer will give companies a centralized vision of their position of AI, which allows them to administer the risks and threats to detect before they become a concern.
“As the use of AI increases, security remains a main concern, we often hear that organizations are concerned about the risks that can come with a rapid adoption,” said Archana Ramamoorthy, senior director of product management, Google cloud security. “Google Cloud undertakes to help our clients build and implement with AI in a safe, compatible and private way.”
Increased protection for AI workloads
AI Protection will be integrated into the Security Command Center (SCC), providing a centralized IA security management system along with other cloud risks.
Among the central capabilities of the new platform are the discovery of AI (identifies and evaluates AI assets for vulnerabilities), the safety of AI assets (implements, policies and railings to ensure AI resources) and threat management (it offers detection, research and response mechanisms for threats related to AI).
In addition, Google Cloud explained that its confidential data protection improvements (SDP) are now extended to AI VERTEX data sets, allowing automatic discovery and classification of confidential training and adjustment data. After discovering confidential data, AI Protection will use the SCC Virtual Red Equipment to identify possible attack routes in AI systems and suggest remediation steps.
Google Cloud also said that Model Armor, a central capacity for AI protection, is now generally available. It is designed to protect against injection and jailbreak attacks, malicious data loss and URLs, and offensive content. It can be integrated into applications through Rest API, Apigee and soon vertex ai.
Finally, the protection of AI will operationalize the intelligence of security and research of Google and Mandiant to help defend the AI systems.
Initial access attempts, privilege escalation and persistence attempts for AI workloads can be detected through SCC, while new detectors for AI protection, based on the last first -line intelligence, “will arrive soon.” These will help identify and administer threats of execution time, such as the kidnapping of fundamental models.
