Stage 9: AI Security, Blockchain, and ZKML

9 Secure, govern, and verify AI-enabled systems.

Goal

Learn AI and agent security, application security, governance, blockchain fundamentals, smart contract risk, zero-knowledge concepts, and practical ZKML limits.

Roadmap to Master This Stage

1. Read the stage goal and diagram before opening the parts.
2. Move through the parts in order unless you can already pass the exit criteria.
3. Study each sub-part folder: overview, deep dive, and examples/practice.
4. Build the stage artifact in small slices and measure the listed metrics.
5. Use the part exam after each part, or open the global Exam tab to test across the roadmap.

Stage Structure Diagram

%%{init: {"flowchart": {"htmlLabels": true, "nodeSpacing": 70, "rankSpacing": 90}, "themeVariables": {"fontSize": "18px"}} }%%
flowchart LR
  P1["<b>9.1</b><br/>LLM and Agent Security"]
  P2["<b>9.2</b><br/>Secure AI Application<br/>Architecture"]
  P1 --> P2
  P3["<b>9.3</b><br/>Governance and<br/>Responsible AI"]
  P2 --> P3
  P4["<b>9.4</b><br/>Blockchain Fundamentals"]
  P3 --> P4
  P5["<b>9.5</b><br/>Smart Contract Security"]
  P4 --> P5
  P6["<b>9.6</b><br/>ZK and Verifiable AI"]
  P5 --> P6

Parts

Part	Simple explanation	Build focus
9.1 LLM and Agent Security	Protect systems where models read untrusted content and call tools.	Threat-model the Stage 6 agent.
9.2 Secure AI Application Architecture	Apply ordinary AppSec and privacy controls to AI services and data flows.	Create an AI risk register and controls plan.
9.3 Governance and Responsible AI	Map risks to controls, monitoring, human oversight, documentation, and review processes.	Write a governance note for a high-impact AI feature.
9.4 Blockchain Fundamentals	Understand wallets, signatures, transactions, gas, state, and finality before AI systems touch chain state.	Trace a simple transaction workflow.
9.5 Smart Contract Security	Recognize common contract vulnerabilities and design AI-to-chain permissions safely.	Create vulnerable contracts, exploits, and fixes.
9.6 ZK and Verifiable AI	Use cryptographic verification for trust, privacy, and constrained model claims.	Prove a tiny model-related computation.

Sub-Part Map

Part	Sub-part	Why it matters
9.1	9.1.1 Prompt Injection and Jailbreaks	Prompt Injection and Jailbreaks is the working skill inside LLM and Agent Security that helps you build the stage artifact, A threat model, red-team report, smart contract security lab, and tiny ZKML or verifiable computation demo, while collecting enough evidence to trust the result.
9.1	9.1.2 Tool Injection and Excessive Agency	Tool Injection and Excessive Agency is the working skill inside LLM and Agent Security that helps you build the stage artifact, A threat model, red-team report, smart contract security lab, and tiny ZKML or verifiable computation demo, while collecting enough evidence to trust the result.
9.1	9.1.3 Data Leakage and Prompt Logs	Data Leakage and Prompt Logs is the working skill inside LLM and Agent Security that helps you build the stage artifact, A threat model, red-team report, smart contract security lab, and tiny ZKML or verifiable computation demo, while collecting enough evidence to trust the result.
9.1	9.1.4 Insecure Output Handling	Insecure Output Handling is the working skill inside LLM and Agent Security that helps you build the stage artifact, A threat model, red-team report, smart contract security lab, and tiny ZKML or verifiable computation demo, while collecting enough evidence to trust the result.
9.2	9.2.1 Authentication and Authorization	Authentication and Authorization is the working skill inside Secure AI Application Architecture that helps you build the stage artifact, A threat model, red-team report, smart contract security lab, and tiny ZKML or verifiable computation demo, while collecting enough evidence to trust the result.
9.2	9.2.2 Secrets and Service Accounts	Secrets and Service Accounts is the working skill inside Secure AI Application Architecture that helps you build the stage artifact, A threat model, red-team report, smart contract security lab, and tiny ZKML or verifiable computation demo, while collecting enough evidence to trust the result.
9.2	9.2.3 Dependency and Model Supply Chain	Dependency and Model Supply Chain is the working skill inside Secure AI Application Architecture that helps you build the stage artifact, A threat model, red-team report, smart contract security lab, and tiny ZKML or verifiable computation demo, while collecting enough evidence to trust the result.
9.2	9.2.4 Privacy PII and Data Retention	Privacy PII and Data Retention is the working skill inside Secure AI Application Architecture that helps you build the stage artifact, A threat model, red-team report, smart contract security lab, and tiny ZKML or verifiable computation demo, while collecting enough evidence to trust the result.
9.3	9.3.1 Risk Mapping	Risk Mapping is the working skill inside Governance and Responsible AI that helps you build the stage artifact, A threat model, red-team report, smart contract security lab, and tiny ZKML or verifiable computation demo, while collecting enough evidence to trust the result.
9.3	9.3.2 Human Oversight	Human Oversight is the working skill inside Governance and Responsible AI that helps you build the stage artifact, A threat model, red-team report, smart contract security lab, and tiny ZKML or verifiable computation demo, while collecting enough evidence to trust the result.
9.3	9.3.3 Fairness Bias and Toxicity	Fairness Bias and Toxicity is the working skill inside Governance and Responsible AI that helps you build the stage artifact, A threat model, red-team report, smart contract security lab, and tiny ZKML or verifiable computation demo, while collecting enough evidence to trust the result.
9.3	9.3.4 Documentation and Auditability	Documentation and Auditability is the working skill inside Governance and Responsible AI that helps you build the stage artifact, A threat model, red-team report, smart contract security lab, and tiny ZKML or verifiable computation demo, while collecting enough evidence to trust the result.
9.4	9.4.1 Wallets Keys and Signatures	Wallets Keys and Signatures is the working skill inside Blockchain Fundamentals that helps you build the stage artifact, A threat model, red-team report, smart contract security lab, and tiny ZKML or verifiable computation demo, while collecting enough evidence to trust the result.
9.4	9.4.2 Transactions Blocks and Gas	Transactions Blocks and Gas is the working skill inside Blockchain Fundamentals that helps you build the stage artifact, A threat model, red-team report, smart contract security lab, and tiny ZKML or verifiable computation demo, while collecting enough evidence to trust the result.
9.4	9.4.3 Smart Contract State	Smart Contract State is the working skill inside Blockchain Fundamentals that helps you build the stage artifact, A threat model, red-team report, smart contract security lab, and tiny ZKML or verifiable computation demo, while collecting enough evidence to trust the result.
9.4	9.4.4 Oracles and Off Chain Data	Oracles and Off Chain Data is the working skill inside Blockchain Fundamentals that helps you build the stage artifact, A threat model, red-team report, smart contract security lab, and tiny ZKML or verifiable computation demo, while collecting enough evidence to trust the result.
9.5	9.5.1 Reentrancy and External Calls	Reentrancy and External Calls is the working skill inside Smart Contract Security that helps you build the stage artifact, A threat model, red-team report, smart contract security lab, and tiny ZKML or verifiable computation demo, while collecting enough evidence to trust the result.
9.5	9.5.2 Access Control Bugs	Access Control Bugs is the working skill inside Smart Contract Security that helps you build the stage artifact, A threat model, red-team report, smart contract security lab, and tiny ZKML or verifiable computation demo, while collecting enough evidence to trust the result.
9.5	9.5.3 MEV Front Running and Oracle Risk	MEV Front Running and Oracle Risk is the working skill inside Smart Contract Security that helps you build the stage artifact, A threat model, red-team report, smart contract security lab, and tiny ZKML or verifiable computation demo, while collecting enough evidence to trust the result.
9.6	9.6.1 Zero Knowledge Fundamentals	Zero Knowledge Fundamentals is the working skill inside ZK and Verifiable AI that helps you build the stage artifact, A threat model, red-team report, smart contract security lab, and tiny ZKML or verifiable computation demo, while collecting enough evidence to trust the result.
9.6	9.6.2 Circuits Witnesses and Provers	Circuits Witnesses and Provers is the working skill inside ZK and Verifiable AI that helps you build the stage artifact, A threat model, red-team report, smart contract security lab, and tiny ZKML or verifiable computation demo, while collecting enough evidence to trust the result.
9.6	9.6.3 zkVMs and Fixed Point ML	zkVMs and Fixed Point ML is the working skill inside ZK and Verifiable AI that helps you build the stage artifact, A threat model, red-team report, smart contract security lab, and tiny ZKML or verifiable computation demo, while collecting enough evidence to trust the result.
9.6	9.6.4 ZKML Use Cases and Limits	ZKML Use Cases and Limits is the working skill inside ZK and Verifiable AI that helps you build the stage artifact, A threat model, red-team report, smart contract security lab, and tiny ZKML or verifiable computation demo, while collecting enough evidence to trust the result.

Stage Artifact

A threat model, red-team report, smart contract security lab, and tiny ZKML or verifiable computation demo.

What to Measure

• security tests
• attack success before and after mitigation
• risk severity table
• proof generation time
• verifier time or gas
• model size limits

Exit Criteria

• threat-model LLM and agent systems
• apply least privilege
• explain smart contract risks
• build or explain a tiny ZK proof and limits

Navigation

Previous: Stage 8: Optimization and Hardware Acceleration | Next: Stage 10: Mastery