Artificial Intelligence & Machine Learning , Events , Next-Generation Technologies & Secure Development

The New Investment Frontier: Defending AI Models, Algorithms

Artificial intelligence and machine-learning technology is vulnerable to cyberattacks due to a lack of security around the models themselves, said Mark Hatfield, founder and general partner, Ten Eleven Ventures.

See Also: Safeguarding Election Integrity in the Digital Age

From predictive analytics that bridges the basic gap in cybersecurity posture to the use of artificial intelligence technology in a myriad of threat intelligence applications, Hatfield said AI and ML algorithms are constantly in use. But what happens if AI systems get attacked? How do we identify and fix the potential risks of misuse that come with AI?

Hatfield said the big driver of risk is inadequate investment in security. The vulnerability this creates extends from algorithms used by companies such as Facebook to insurance brokers that use these models to underwrite health and cyber insurance.

"We've been doing investments in companies that are leveraging AI and machine learning into their solutions," Hatfield said. "While it's not new from a cyber perspective, I think the twist that's different is now, 'How do we protect what has been built from the model's perspective?' That's a new aspect."

In this video interview with Information Security Media Group at RSA Conference 2023, Hatfield also discusses:

• The difference between defending an AI model versus a network or endpoint;
• The implications of incorporating generative AI at scale for cyber defenders;
• Why Ten Eleven has stayed away from startups focused on offensive cyber.

Hatfield previously was a partner at Fairhaven Capital and managing director for Motorola Ventures. He has also held various corporate business development and IT roles at Bell Canada Enterprises and the Bank of Montreal.