Understanding AI Risks and Exposures
The specific ways AI systems fail, and how each failure turns into insurable exposure. Work through bias, hallucination, drift, security and autonomy, then follow each one to the line of business it lands on, across cyber, liability, property and motor.
Turn "AI went wrong" into a named exposure on a named line
This course assumes you already know AI is a risk and takes you to the next step: naming exactly how a given system fails, and where that failure attaches to cover. Every failure mode is worked through with real loss scenarios, then mapped onto the policies that respond and the gaps that do not.
- Classify an AI failure by mode, whether it is bias, hallucination, drift, a security compromise or an autonomy error, and describe the loss each one produces.
- Trace each failure to the line of business it lands on, across cyber, professional and general liability, property and motor, including where two lines both plausibly respond.
- Spot silent AI exposure sitting inside policies that were never written with AI in mind, and flag it before it surfaces as a disputed claim.
- Rank a book or a submission by which AI failure modes carry the most severe and most likely loss, so attention goes where the exposure actually is.
Six lessons, about six hours
Self-paced and on demand. Each lesson ends with a short knowledge check; a final assessment, earns the certificate of completion.
01 A working taxonomy of AI failure
The five failure modes that matter for insurance, why they are distinct, and the shared vocabulary you will use to place any AI incident before you reason about cover.
02 Bias and discriminatory outcomes
How a model learns and reproduces unfair outcomes, the third-party harm that follows, and the liability, employment practices and management liability exposure it creates.
03 Hallucination and fabricated output
Why generative systems produce confident, wrong answers, the professional-services and media losses that result, and where errors and omissions cover does and does not reach.
04 Drift and silent degradation
How a model that worked at deployment quietly decays as the world moves under it, why the loss is gradual and hard to date, and what that means for triggers and aggregation.
05 Security failures unique to AI
Data poisoning, prompt injection, model theft and adversarial inputs: the attack surface AI adds on top of ordinary cyber risk, and how it reads against a cyber wording.
06 Autonomy, physical action, and mapping it all to cover
Where an AI system acts in the physical world and creates property and motor exposure, then a full mapping of every failure mode to the lines that respond, and the final assessment.
Built for the desk that has to place the exposure, not just name it
You already accept that AI carries risk. What you need is a reliable way to move from a vague concern to a specific exposure on a specific line, so you can underwrite it, place it or reserve for it. A grounding equivalent to AI Foundations helps, but the course reintroduces each failure mode as it goes.
Underwriters
You price accounts touched by AI and need each failure mode tied to the line it threatens before you set terms.
Brokers and account teams
Your clients are deploying AI across functions. You want to show them exactly where the exposure sits and which cover answers it.
Claims and risk managers
You have to recognise which failure mode drove a loss, and reason about which policy should respond and which will contest it.
One core course in a larger certification
Understanding AI Risks and Exposures is one of the 4 core courses that make up the CAIIP certification, the Certified AI Insurance Practitioner. Complete all of them, choose one specialization, and pass the proctored exam to become certified. CAIIP is coming 2026.
AI Foundations
View courseUnderstanding AI Risks and Exposures
Governing and Managing AI Risk
View courseAssessing and Insuring AI Exposure
View courseAI Risk in Cyber Insurance
View courseAI Risk in Liability Insurance
View courseAI Risk in Property Insurance
View courseAI Risk in Motor and Autonomous Systems
View courseAbout this certificate course
No. Understanding AI Risks and Exposures is a certificate course, which is shorter than a certification. You earn a certificate of completion, not a certification credential.
It helps but it is not required. This course is pitched at intermediate level and reintroduces each failure mode as it goes, so you can take it on its own. If you are new to AI, AI Foundations is the natural starting point.
There is a final assessment. Each lesson has a short knowledge check along the way, and the final assessment confirms you are ready for the certificate of completion.
No. It is a one-off purchase. There is no renewal and no maintenance fee. Your certificate of completion does not expire.
Yes. Understanding AI Risks and Exposures is one of the four core courses of the CAIIP certification, the Certified AI Insurance Practitioner, which is coming in 2026. Taking it on its own still earns the standalone certificate of completion.
Name the failure, find the line
About six hours to move from "AI is a risk" to a named exposure on a named line, across cyber, liability, property and motor. A certificate course, and a CAIIP core course.