Artificial Intelligence Conference - Virtual-Course-Handout - AI 201 Data Security, Privacy, Security On AI

Course-Handout - AI 201 Data Security, Privacy, Security On AI

Pdf Summary

This presentation on AI security and privacy explains the major risks businesses face when using generative AI tools and how to reduce them. Key risks include data privacy breaches, data security exposure, hallucinations, biased outputs, copyright infringement, and deepfakes. It emphasizes that users should never upload sensitive or private information into public AI platforms and should scrub identifying details such as names, SSNs, and account numbers before using AI.<br /><br />The session reviews copyright issues, noting that AI-only output may be treated as public domain in some jurisdictions, while AI-assisted work may be copyrightable if there is meaningful human input. It also explains that training AI on copyrighted data remains a legally unsettled area, with fair use depending on factors like transformativeness, amount used, and market harm.<br /><br />A major theme is understanding platform terms of use, privacy policies, and acceptable use policies. Users should know what data AI companies collect, how that data may be used, and whether they can opt out of model training. The presentation compares policies and protections across major tools such as ChatGPT, Gemini, Copilot, Claude, and Perplexity.<br /><br />It also surveys privacy and AI regulations in the EU, Canada, and many U.S. states, highlighting laws such as GDPR, the EU AI Act, state privacy statutes, and recent AI-specific legislation in California and Utah.<br /><br />Finally, it describes three core GenAI risk categories: input risk, output risk, and system risk, and warns about deepfake scams that impersonate executives and enable fraud. The overall advice is to manage AI settings carefully, limit data exposure, train employees, and treat AI use as both a security and compliance issue.

Keywords

AI security

data privacy

generative AI risks

hallucinations

deepfakes

privacy policies

model training

GDPR

AI compliance