Seattle cybersecurity startup Protect AI landed $35 million to boost the rollout of its platform that helps enterprises shore up their machine learning code.
Protect AI sells software that allows companies to monitor the various layers and components of machine learning systems, detecting potential violations and logging information on those attacks. It primarily sells to large enterprises in regulated industries including finance, healthcare, life sciences, energy, government, and tech.
The fresh funding comes as AI has become a focal point for many enterprise-level executives, who are mandated to deploy the tech alongside their product suites, CEO Ian Swanson told GeekWire. This rapid adoption comes with elevated risks, he said.
“[AI] is flying down the highway right now,” he said. “For a lot of organizations, that can’t be stopped. So we need to make sure that we can maintain and understand it.”
A KPMG survey found than only 6% of organizations have a “dedicated team in place for evaluating risk and implementing risk mitigation strategies as part of their overall generative AI strategy.”
At the same time, companies of all sizes are facing an increasing number of cyber threats, pressuring execs to invest heavily in their security systems. McKinsey and Co. predicts businesses will spend more than $100 billion on related services by 2025.
Protect AI’s flagship product, AI Radar, creates a machine learning bill of materials to track a company’s “software supply chain” components: operations tools, platforms, models, data, services, and cloud infrastructure. Swanson compares it to regular automotive maintenance and inspection, where tires and brakes need constant checks, along with ensuring the right fuel is used.
“We really have to understand the ingredients and the recipe of all this,” he said.
A hacker gaining access to a company’s machine learning system can steal intellectual property or inject malicious code, Swanson said. For instance, Protect AI found a vulnerability in MLflow, a popular machine learning lifecycle platform used by Walmart, Time Warner, Prudential, and other large companies.
The startup presented its findings in March, pressuring MLflow to update its platform within a few weeks. The flaw, left unpatched, would have allowed unauthenticated hackers to read any file accessible on a user’s MLflow server and potentially inject code.
Protect AI’s first product was NB Defense, an open-sourced app that works to address vulnerabilities in development platform Jupyter Notebooks. Protect AI’s tools work in Google Cloud, Oracle Cloud, Microsoft Azure and Amazon Web Services.
In the AI cybersecurity space, there are several well-funded startups.
- Hidden Layer raised $6 million in funding last year for its tool that detects attacks on AI models.
- Robust Intelligence landed $30 million in 2021 for its product that stress tests AI models.
- CalypsoAI recently nabbed $23 million for its tool that validates and monitors AI apps before they are developed.
Swanson said Protect AI tracks the entire machine learning supply chain, from the original inputted training sets to the ongoing use of the model.
This is Swanson’s third startup. His first company was Sometrics, a virtual currency platform and in-game payments provider. It was acquired by American Express in 2011. After that, he founded DataScience.com, a cloud workspace platform that was acquired by Oracle in 2018. Swanson also held AI leadership roles at AWS and Oracle.
Swanson is joined by Badar Ahmed, a former engineering leader at Oracle and DataScience, and Daryan Dehghanpisheh, a former leader at AWS. The company has 25 employees, up from 15 when the company raised its $13.5 million seed round in December.
The Series A round was led by Evolution Equity Partners, with participation from Salesforce Ventures and existing investors Acrew Capital, Boldstart Ventures, Knollwood Capital, and Pelion Ventures. The startup has raised a total of $48.5 million to date.
