Examples of the Different AI Risks
Pete Hannam
Data and AI Systems Architect
To understand the different tiers ot AI risk, look at some examples of unacceptable risk, high risk, limited risk and minimal risk AI.
To understand the different tiers ot AI risk, look at some examples of unacceptable risk, high risk, limited risk and minimal risk AI.
Subscribe to watch
Access this and all of the content on our platform by signing up for a 7-day free trial.
Examples of the Different AI Risks
53 secs
Key learning objectives:
Name examples for each tier of AI risk
Overview:
AI risk can vary widely. That's why some AI systems are completely banned (e.g. anything that deliberately exploits the vulnerable or involves social scoring) and some require little to no oversight (inbox spam filters). Some in the middle (AI involving medical systems, the justice system, law enforcement and transport) can operate but only after meeting stringent requirements.
Subscribe to watch
Access this and all of the content on our platform by signing up for a 7-day free trial.
Summary
What are some examples of high risk AI?
Pathway 1: For AI integrated into products
Pathway 2: For standalone AI
What are some examples of limited risk AI?
What are some examples of minimal risk AI?
What are some examples of unacceptable risk AI?
- Subliminal manipulation
- Exploitation of vulnerabilities
- Social scoring
- Expansion of facial recognition databases through untargeted scraping
- Emotion recognition in specific contexts (e.g. workplaces)
- Individual predictive policing
What are some examples of high risk AI?
Pathway 1: For AI integrated into products
- Medical devices and in-vitro diagnostics
- Machinery and equipment
- Toys
- Vehicles and aviation systems
- Marine equipment
- Rail systems
- Personal protective equipment
Pathway 2: For standalone AI
- Biometric systems
- Critical infrastructure
- Education and vocational training
- Employment and worker management
- Essential services
- Law enforcement
- Migration, asylum and border control
- Justice
What are some examples of limited risk AI?
- Chatbots and conversational AI
- Deepfakes
- AI-generated text on matters of public interest
- Emotion recognition (e.g. call centres)
What are some examples of minimal risk AI?
- Spam filters protecting your inbox
- Recommendation engines for films or music
- Route optimisation for deliveries
- Inventory management systems
- Translation tools
- Photo enhancement features
Subscribe to watch
Access this and all of the content on our platform by signing up for a 7-day free trial.
Pete Hannam
Pete Hannam has over twenty years’ experience building large-scale data platforms, including those that now underpin modern AI systems. His work has ranged from national security infrastructure processing hundreds of billions of records to global enterprise platforms supporting tens of thousands of users. He has architected AI-ready systems for government, consultancies, and enterprises, with a strong focus on governance, security, and regulatory compliance. His work emphasises that successful AI deployment depends not only on technology, but on understanding the organisational, technical, and regulatory context that enables AI initiatives to deliver real value.
There are no available Videos from "Pete Hannam"