Cybersecurity for machine learning models startup HiddenLayer Inc. today announced the public launch of a new artificial intelligence cybersecurity platform that empowers security operations and data science teams to protect their AI investments against adversarial machine learning attacks.

The HiddenLayerMLSec Platform is comprised of the company’s flagship products HiddenLayer MLDR, ModelScanner and Security Audit Reporting. HiddenLayer MLDR uses a machine learning-based approach to analyze model events in real-time to identify malicious activity without requiring access to the organization’s models or sensitive training data.

The company argues that adversarial machine learning attacks are the newest and most destructive development of the threat landscape, targeting advanced and important technology. The attacks are said to be an evolution of traditional cybersecurity threats such as malware, ransomware, phishing, spam and data breaches, and modern environments are not prepared to defend against them.

The parts of the platform include real-time defense, flexible response options and configurable settings that are fine-tuned to a company’s needs. The ModelScanner scans and guarantees model integrity by identifying vulnerabilities and ensuring models have not been compromised through malicious code injections.

Security Audit Reporting validates machine learning model security across the enterprise, providing a comprehensive view of the security status of AI and machine learning assets. An on-demand dashboard and distributable reporting are accompanied by vulnerability prioritization.

“The rapid adoption of AI/ML introduces a new attack surface for threat actors to exploit, so we need the tools and knowledge to be able to shed light into the areas our machine learning models are most vulnerable to attack,” explained Malcolm Harkins, chief security and trust officer at Epiphany Systems Inc., one of HiddenLayer’s existing customers. “The HiddenLayer team consists of subject matter experts in the domain of artificial intelligence and cybersecurity. They have developed products that allow our Security Operations teams to stay ahead of the new threats in AI.”

In addition to the announcement of the new platform, HiddenLayer is also seeking early design partners looking to fulfill their cybersecurity needs as they increase their investments in AI and want to do it safely.

HiddenLayer is a venture capital-backed startup that has raised $6 million to date. According to Crunchbase, investors in the company include Ten Eleven Ventures.

Image: HiddenLayer