SecAI+ Cloud vs On Premises AI Deployment Security

1. In cloud AI environments, what does VPC (Virtual Private Cloud) isolation primarily protect against?

Unauthorized access from other cloud tenants

Model overfitting

Inference latency issues

Training data imbalance

2. In comparing cloud and on-premises AI security, which statement about vendor lock-in is most accurate?

Cloud deployments carry higher vendor lock-in risk; on-premises offer more portability

On-premises deployments have more vendor lock-in

Both have identical lock-in risks

Vendor lock-in is irrelevant to security

3. What is the primary security benefit of using secure enclaves or trusted execution environments in AI deployments?

Protects model inference from unauthorized access even by privileged users

Increases model accuracy

Reduces memory requirements

Improves network bandwidth

4. Which security control is most critical for protecting against unauthorized model deployment in on-premises environments?

Access control and change management processes

High-speed networking

Unlimited storage capacity

Public model repositories

5. In cloud AI deployments, what is the purpose of audit logging and monitoring?

Track user actions, detect anomalies, and ensure accountability

Increase model training speed

Reduce cloud storage requirements

Eliminate the need for backups

6. What security challenge arises from model extraction attacks in both deployment models?

Attackers reverse-engineer proprietary models through API queries

Models automatically delete themselves

Training data becomes public

Servers lose internet connectivity

7. Which encryption approach is essential for protecting AI training data in cloud environments?

Encryption at rest and in transit with key management

Encryption only during transmission

Encryption only at storage

No encryption if using HTTPS

8. In on-premises deployments, what does network segmentation using firewalls primarily achieve?

Restricts unauthorized access between network zones

Increases model inference speed

Reduces power consumption

Improves training accuracy

9. What is the primary security risk of deploying AI models without API rate limiting in cloud environments?

Denial-of-service attacks and resource exhaustion

Model accuracy degradation

Slower model training

Increased storage costs only

10. Which security audit requirement is typically stricter for on-premises AI deployments in regulated industries?

Physical security audits and compliance certifications

Code formatting standards

API naming conventions

Documentation color schemes

11. Which security advantage does on-premises AI deployment offer over cloud?

Full physical control and data residency

Automatic security patching from vendors

Lower total cost of ownership

Built-in multi-region redundancy

12. What is the primary security advantage of containerization in both cloud and on-premises AI deployments?

Isolates application dependencies and reduces attack surface

Eliminates the need for firewalls

Increases model training speed

Removes encryption requirements

13. Which cloud AI security concern relates to multiple tenants sharing underlying hardware?

Side-channel attacks and resource contention

Insufficient CPU cores for inference

Inability to use GPUs

Lack of model versioning

14. In on-premises AI security, what does data residency requirement mean?

Data must remain within specific geographic or jurisdictional boundaries

Data must be encrypted with symmetric algorithms

Data must be replicated across multiple cloud regions

Data must be deleted after 90 days

15. What is the primary purpose of identity and access management (IAM) in cloud AI deployments?

Control who accesses AI models, data, and infrastructure resources

Increase model training speed

Reduce cloud storage costs

Improve inference accuracy

16. Which threat is most prominent in on-premises AI deployments with insufficient network segmentation?

Lateral movement and internal lateral privilege escalation

Cloud provider data breaches

Internet routing attacks

DNS hijacking from external sources

17. What security benefit do managed cloud AI services provide through shared responsibility models?

Vendor handles infrastructure security; organization handles data and access control

Cloud provider assumes all security responsibility

Organization has no security obligations

Both parties are equally responsible for all layers

18. Which on-premises deployment security measure is most critical for protecting proprietary AI models?

Air-gapped network architecture

High-speed internet connectivity

Public API exposure

Automatic cloud backups

19. In a cloud AI deployment, what does encryption in transit primarily protect against?

Network interception and man-in-the-middle attacks

GPU memory corruption

Model inference latency

Cloud provider billing errors

20. What is a primary compliance risk when deploying AI models in public cloud environments?

Data sovereignty and regulatory jurisdiction issues

Inability to scale inference workloads

Lack of GPU availability

Incompatibility with open-source frameworks

SecAI+ Cloud vs On Premises AI Deployment Security Quiz