Artificial intelligence and machine learning cybersecurity company Protect AI Inc. today unveiled Sightline, a vulnerability database that provides insights into known and emerging AI and machine learning  vulnerabilities along with an early warning system to defend against threats.

Designed to enable organizations to detect, assess and remediate their AI ecosystem and environment quickly for associated AI risks, Sightline also allows businesses to embrace AI and machine learning models with confidence and safety.

The rise of AI and machine learning has been well-documented, but only recently have the supply chain risks in the libraries, packages, frameworks, foundational models and datasets used in AI gained more attention. Protect AI argues that while the supply chain is the backbone of much AI innovation and digital transformation, existing vulnerability databases lack proper AI coverage, fail to provide necessary details on AI risks and don’t address all the complexities of securing AI applications and machine learning models. This is where Sightline steps.

“Traditional vulnerability databases fall short, a gap that Sightline uniquely addresses,” Ian Swanson, chief executive officer of Protect AI, explains. “Sightline doesn’t just add to the vulnerability database noise; it zeroes in on AI and ML vulnerabilities with precision, backed by our exclusively focused AI/ML hacker community — huntr.”

Protect AI”s “huntr” bug bounty community is claimed to be the industry’s only dedicated AI and machine learning hacker community, with over 15,000 experts uncovering vulnerabilities in under-researched areas and foundational AI components. Sightline taps into the insights uncovered by huntr to ensure that enterprises stay ahead in securing their AI applications.

The huntr bug bounty community is currently uncovering and fixing more than 50 unique vulnerabilities in the AI/machine learning supply chain monthly. Previous notable finds include issues with Remote Code Executions in Hugging Face Transformers, Nvidia Corp.’s Triton Interface Server and PyTorch Serve.

Also included with Sightline are penetration testing tools that can quickly assess vulnerabilities within AI environments and provide additional context on the impact of issues. Sightline integrates with Protect AI’s Radar security posture management product to contextualize vulnerabilities within a customer’s AI supply chain. That allows for swift and accurate responses to potential threats by automatically evaluating the relevance of vulnerabilities within the unique context of their AI applications.

Sightline joins Protect AI’s growing suite of AI security tools, including Guardian, a secure gateway launched in January that helps organizations enforce security policies on machine learning models to prevent malicious code from entering their environment.

Image: Protect AI