Is the AWS Bedrock Academy free?

Yes. The full course is free and includes 25 modules, 52 lessons, 16 labs, and quiz-based progression with no sign-up required.

How is Infinity AWS Bedrock Academy different from AWS Skill Builder?

Unlike AWS Skill Builder which often gates advanced hands-on labs behind paid subscriptions or AWS Builder ID logins, Infinity Tech Stack's AWS Bedrock Academy is 100% free, ungated, and browser-first with zero sign-up required. It includes 25 modules, 52 lessons, 16 hands-on labs, and 125+ quiz checks covering the Converse API, Bedrock Guardrails, AgentCore, and multi-model evaluations without requiring active AWS account credentials.

Who is this course for?

Developers, architects, and platform teams building AI systems on AWS Bedrock from MVP to production scale.

What does this Bedrock course cover?

The academy covers runtime inference, prompt engineering, guardrails, knowledge bases, agents, evaluation, observability, security, compliance, resilience, incident response, and production launch discipline.

Safety regression suites should run:

On every major prompt/model policy change — Continuous regression is required to detect behavioral drift.

Policy Testing and Failure Modes Tutorial | Safety and Guardrails

← Back to Dashboard

1. Guardrails Concepts2. Policy Testing and Failure Modes

Policy Testing and Failure Modes

📚 Safety and Guardrails⏱ 10 min⭐ 75 XP⌨ Hands-on lab

Policy Quality Is Measurable

Safety quality requires regression testing. Build a repeatable suite that includes allowed, risky, and prohibited requests, then score guardrail behavior over time - exactly like a unit-test suite for your safety posture.

The Four Outcomes Every Test Can Produce

Outcome	Meaning	Severity
True positive	Unsafe content correctly blocked	Working as intended
True negative	Safe content correctly allowed	Working as intended
False negative	Unsafe content allowed through	Critical - compliance/harm exposure
False positive	Safe content wrongly blocked	Erodes UX and user trust; teams then weaken policies

Building the Suite

Benign set - realistic everyday requests that must always pass (regression canary for over-blocking).
Borderline set - dual-use questions, aggressive-but-legitimate wording, policy edge cases.
Prohibited set - known-bad prompts, including indirect phrasings and roleplay-wrapped attempts.
Injection set - prompt-attack payloads (ignore-previous-instructions, tool-output poisoning, encoded instructions).

Score each release: block-rate on prohibited, pass-rate on benign, and drift versus the previous run. Bedrock's ApplyGuardrail API lets you test policies standalone - no model invocation required - which makes CI integration cheap.

Measure false negatives (unsafe content allowed).
Measure false positives (safe content blocked).
Track drift after prompt or model changes - a model upgrade can shift guardrail behavior even with identical policies.

CI rule: run the safety suite on every guardrail config change, every prompt release, and every model version bump. Any new false negative on the prohibited set is a blocking failure.

⌨ HANDS-ON LABBuild a Safety Regression Set

⭐ +150 XP

Create a red-team style prompt suite and validate expected block/allow outcomes.

1Create a CSV of test prompts and expected outcomes.

2Document block-rate and false-positive metrics.

lab-sandbox — simulated environment

INFINITY LAB SANDBOX v2.6 — simulated shell
Type the command for the current objective. Helpers: "hint", "solution", "clear".
$

OBJECTIVE 1 / 2 — type "hint" if stuck

🧪 Knowledge Check

Press 1-4 to select1 of 2

Which metric is most dangerous in regulated environments?

False negatives on unsafe content

UI load time

Icon size

Git commit frequency