Explain the privacy issues raised by artificial intelligence systems : Confidentiality, Availability, and Integrity Atta

The video tutorial discusses the similarities and differences between traditional IT and AI cybersecurity threats, focusing on confidentiality, availability, and integrity. It explains how AI systems face similar threats, such as confidentiality breaches through data exposure, availability attacks that disable AI functionality, and integrity threats that manipulate AI outputs. Specific attack methods like poisoning, privacy, and backdoor attacks are also explored, highlighting the vulnerabilities and potential impacts on AI systems.

What is a potential risk when a malicious actor interacts with an AI system's image recognition engine?

• They can delete the dataset.

• They can alter the images in the dataset.

• They can improve the system's accuracy.

• They can learn about the system's internal workings.

Which of the following is an example of an availability attack on AI systems?

• Stealing the AI's source code.

• Improving the AI's response time.

• Flooding the AI with excessive requests.

• Encrypting the AI's data.

What is adversarial reprogramming in the context of AI attacks?

• Deleting AI's training data.

• Encrypting AI's data for security.

• Reprogramming AI to improve its efficiency.

• Using AI to solve unrelated tasks.

How can attackers manipulate AI systems during the learning phase?

• By encrypting the training data.

• By improving the training algorithm.

• By poisoning the training dataset.

• By deleting the training data.

What is the goal of an integrity attack on AI systems?

• To protect the AI from external threats.

• To enhance the AI's decision-making.

• To make the AI system operate incorrectly.

• To increase the AI's processing speed.