Secure development practices
AI Security and Risk Management
Angeline Corvaglia
Founder and Digital Transformation Expert
Counteract the unique vulnerabilities of AI
Data provenance and traceability
- Data provenance: origin of data
- Traceability: tracking its journey
How it helps security
- Data poisoning attacks
- Corrupted data
Validation and sanitation
- Checking data
- Cleaning data
Helps ensure data is:
- accurate
- relevant
- free from any malicious tampering
Data minimization
- Limiting the amount of data collected and processed
- Only what is absolutely necessary
Helps reduce the risk of:
- privacy breaches
- data theft
Encryption and anonymization
- Encryption: Transforms sensitive data into a secure format
- Anonymization: Remove personally identifiable information
How they help security:
- Attacks aimed at extracting sensitive information
Secure coding practices
- Reduce vulnerabilities within the code base
- Regular code reviews
- Vulnerability scanning
- Adhering to coding standards
How it helps security
- Infrastructure attacks
Access controls
- Limiting who can interact
- Trusted, verified individuals or entities
How it helps security
- External attacks exploiting vulnerabilities
Secure infrastructure
- Safeguard both hardware and software
- Firewalls
- Intrusion detection systems
- Secure cloud services
How it helps security
- Infrastructure-targeted threats
Security isn't for the tech team alone
Combined effort ensures that the AI systems are secure, trusted, and reliable.
