Artificial Intelligence & Machine Learning , Events , Leadership & Executive Communication
AI Bias: The Emerging Threat to Sound Business Decisions
VP and CISO Rick Doten on Often Overlooked Risks of Relying on Generative AI
Bias lurks everywhere in generative artificial intelligence: in the data, in the model, in the human interpreting the output of a model. That's why one of the biggest emerging security threats is relying on generative AI for important business decisions, said Rick Doten, vice president for information security at Centene Corp. and CISO of Carolina Complete Health.
See Also: Rethinking Browser Security: From Risk to Asset
"Yes, there's potential for poisoning. Yes, there's potential for putting stuff into the data," he said. "Untrustworthy data making business decisions is the big thing."
Security risks such as prompt injection can be mitigated with guardrails, but generative AI at this point can't tell fact from fiction, leading Doten to advise organizations to develop use cases and assess the risks associated with them.
In this video interview with Information Security Media Group at RSA Conference 2024, Doten discussed:
- Why generative AI is like military intelligence;
- How an unbiased dataset can still produce biased results;
- The importance of guardrails in AI governance.
Doten is a cyber risk management thought leader with more than 30 years of experience in the IT industry, the last 25 focused specifically on cybersecurity. He has worked as both CISO of multinational company and a management consultant performing risk management and risk engineering to mature customer security and privacy programs. He is a member of the CyberEdBoard.
