The guidance provided within this publication focuses on using AI systems securely.
Modern AI systems are among the fastest-growing applications globally. AI is increasingly used to handle activities traditionally undertaken by humans. Examples include:
- sorting large data sets
- automating routine tasks
- creative endeavors
- augmenting customer engagement, logistics, medical diagnosis, and cybersecurity.
While AI can increase efficiency and lower costs, it can also intentionally or inadvertently cause harm to your organization. A few of the challenges include:
- Data Poisoning of an AI Model
- Input manipulation attacks – Prompt injection and adversarial examples
- Generative AI hallucinations
- Privacy and intellectual property concerns
- Model stealing attack
AI technologies are distinctive in their speed of innovation and scope of impact. Therefore, it’s essential for organizations that use or are considering AI systems to evaluate their cybersecurity implications.