HIPAA Compliance Code Scanning

Trace every PHI flow directly in your application's source code to satisfy HIPAA 164.308(a)(1)(ii)(B) risk analysis, 164.312(b) audit controls, and 164.314(a)(2) business associate oversight, before code is ever pushed to production.

Medical History → OpenAI PHI 3P AI RISKY hounddogai/hounddog-test-python-app · 4 steps
  1. 1 Read from the patient record pages/1_Patient_Management.py#L271 271 medical_history = patient.get('medical_history')
  2. 2 Placed inside a string, propagated via str.strip, and assigned to 'patient_context' pages/1_Patient_Management.py#L284-L295 284 patient_context = f""" 285 You are an assistant for clinicians. Use only the provided patient data. 288 - Medical History: {medical_history} 295 """.strip()
  3. 3 Wrapped in langchain_core.messages.SystemMessage and assigned to 'messages' pages/1_Patient_Management.py#L317 317 messages = [SystemMessage(content=patient_context)]
  4. 4 Sent to OpenAI pages/1_Patient_Management.py#L330 330 llm.invoke(messages)

Why HIPAA Compliance Breaks at the Code Level

PHI exposures to risky data sinks like logs, third-party services, and AI integrations are rarely intentional; they happen as codebases grow.

164.308(a)(1)(ii)(B)

Manual Risk Analysis Cannot Keep Up

  • HIPAA requires periodic risk analysis on how PHI is accessed, shared, and stored, but interview-driven assessments lag behind every release

A snapshot of a system that no longer exists.

164.312(b)

Audit Controls Without Code Visibility

  • Current manual workflows and reactive tools infer data flows only after the data has already been flowing, so audit controls cannot tell you where PHI actually lives

No code-level view, no real audit trail.

164.314(a)(2)

BAAs Cannot Police Themselves

  • A signed Business Associate Agreement does not stop a developer from shipping a new PHI field into a third-party SDK

Contract in place, controls missing.

Code in development

Step 1Fetch Patientpatient = get_patient(id)
Step 2Tainted Variableprompt = f"...{patient.history}"
HoundDog.ai intercepts hereTaint-flow analysis catches PHI propagation through every transformation, before code ships to production.
Step 3External Callllm.invoke(prompt)

BAA violation zone

PHI leaks
  • OpenAI
  • Sentry
  • Logs

The result

Stale Risk Documentation
Risk analyses drift the moment a new release ships.
PHI Sprawl in Logs and AI
Medical history and identifiers reach OpenAI, Sentry, and Datadog without anyone signing off.
Penalties Per Violation
HIPAA fines are calculated per record and per category, multiplying fast.

Code-based evidence for HIPAA risk analysis, BAAs & privacy reviews.

At development speed. Catch PHI flows that break Business Associate Agreements before code ships, with privacy teams in control: the engine proposes, the DPO approves.

Discover

Every integration, straight from the code

All third-party and AI integrations detected directly in source code, including Shadow AI, whether the data flows through an SDK or API, with 1,000+ integrations covered out of the box.

  • OpenAI
  • Anthropic
  • LangChain
  • Salesforce
  • Datadog
  • HubSpot
Trace

Follow sensitive data into every sink

Trace 100+ sensitive data types (PII, PHI, CHD, auth tokens) across code paths into logs, storage, APIs, and third-party and AI integrations.

  • LLM Prompts
  • Third-Party SDKs
  • Logs
  • Files
  • Local Storage
  • Many Others
Verify & Suggest

RoPA that keeps itself current

Keep your RoPA updated as new categories of personal data and subprocessors are introduced, detected directly from source code.

Validate design-phase privacy reviews with code-based evidence before code is pushed to production.

Suggest

Org RoPA updates

Verify

Alignment with PIA

Block

Risky data flows

Catch

Log leaks early

How HoundDog.ai Maps to the Three Rules That Trip Up Engineering

Risk analysis, audit controls, and business associate oversight: three Security Rule provisions where the textbook answer breaks once real code ships. Here is what changes when evidence comes from the codebase.

164.308(a)(1)(ii)(B)

Proactive Data Flow Mapping & PHI Inventory

Accurate risk assessment of every ePHI flow across the system.

With HoundDog.ai

  • Source code is the single source of truth for every PHI flow
  • The Datamap inventories every storage, log, and third-party sink that receives PHI
  • Risk analysis updates on every commit, not every quarter
164.312(b)

Proactive Data Minimization & Audit Trail

Mechanisms to record and examine activity in systems that handle ePHI.

With HoundDog.ai

  • PHI in risky sinks is flagged before code is pushed to production
  • PHI storage footprint shrinks to only the locations policy approves
  • Taint-flow analysis traces every PHI element across every code path and transformation, however deep.
164.314(a)(2)

Prevention of Third-Party Oversharing

Business associate contracts and satisfactory assurances PHI is safeguarded.

With HoundDog.ai

  • Every PHI element flowing to each business associate is detected in code
  • Per-vendor allowlists fail the pull request when an unapproved PHI element is added
  • New SDKs and AI services surface as suggested edits before they ship

Typical questions come from different teams

With the self-hosted or cloud platform, your organization’s data map becomes queryable in plain English.

A privacy engineer asks which third-party integrations receive PII.

Which third parties receive PII?

HoundDog.ai Acme Corp
New chat Which third parties receive PII? ···
Ask HoundDog.ai about your data Suggested for you Which repositories handle medical data?Show me the most severe open risky dataflows right nowWhich third parties receive PII?Which AI services receive sensitive data in our code?What data elements are present in my code?
Which third parties receive PII?
Searching…
Searched the datamap
Opened a page
Two third-party data sinks receive PII, counting PHI, PIFI and CHD as PII. Both are in one repository, hounddogai/hounddog-test-python, on the main branch:
Third partyData elementTag and sensitivityDataflows
OpenAI Medical History PHI, high 1 risky dataflow
Sentry Phone Number PII, medium 1 risky dataflow
I've opened the Datamap's Data Sinks view filtered to these two sinks.
Datamap · Data Sinks
Which third parties receive PII? Ask about your data…
8 dataflows 12 dataflows 56 dataflows 43 dataflows 8 dataflows 12 dataflows 56 dataflows 21 dataflows 22 dataflows
53 Data Elements Aadhaar Number PII Auth Token SECRET Bank Account Number PIFI Bank Card Number CHDPIFI Biometrics PHI
HoundDog.ai
7 Data Sinks RISKY Config Files Logs OpenAI 3PAI Sentry 3P SQL Database DB
6 Data Sinks SAFE Discord 3P Files HTTP JSON Web Token OpenTelemetry
1 dataflow 1 dataflow 1 dataflow 1 dataflow
2 Data Elements Medical History PHI Phone Number PII
HoundDog.ai
OpenAI 3PAI RISKY
Sentry 3P RISKY
Without AI

The same Datamap, built from deterministic findings

Without AI, the Datamap is still available, built from the scanner’s deterministic findings. You can apply the same advanced filters to search for dataflows; there is no chat panel and no AI review.

  • By repository — select several, such as backend, frontend, and an authentication service, to see every dataflow across them.
  • By data element — “show me every repository where Social Security Number is processed” and every sink it may be exposed to.
  • By data sink — “show me all sensitive data sent to OpenAI org-wide.”
The Datamap without AI: 53 data elements flowing through HoundDog.ai into 7 risky data sinks, 6 safe data sinks, and 16 data sinks without dataflows, with the Data Sink Tag filter open
The Datamap without AI: 53 data elements flowing through HoundDog.ai into 7 risky data sinks, 6 safe data sinks, and 16 data sinks without dataflows, with the Data Sink Tag filter open

Why not just use an LLM?

HoundDog.ai is built to make your AI more effective, not to replace it.

01

Tokens for reasoning, not discovery.

The scanner discovers every sensitive dataflow, so your AI tokens go to reasoning over detected flows: assessing risk, triaging dataflows and answering questions about your datamap.

02

Evidence you can stand behind.

The same commit always produces the same dataflows. Every engineer, privacy reviewer and agent gets the same answer about where sensitive data goes, instead of output that varies with the prompt.

03

Easy to roll out org-wide.

The Rust scanner runs in CI on standard CPU infrastructure, across every repo and every commit, with negligible impact on build times.

Frequently Asked Questions

How does HoundDog.ai help with HIPAA 164.308(a)(1)(ii)(B) risk analysis?

HoundDog.ai performs static analysis on source code to map every PHI data flow across storage, logs, third-party integrations, and AI services. This produces a continuously updated risk surface that satisfies the periodic risk analysis requirement of HIPAA 164.308(a)(1)(ii)(B) using actual processing behavior, not interview-based estimates.

How does this support HIPAA 164.312(b) audit controls?

HIPAA 164.312(b) requires audit controls over PHI. HoundDog.ai builds a code-level inventory of every place PHI is read, written, and forwarded, and flags PHI reaching unintended sinks like logs or files before code ships. The result is a continuous, evidence-backed trail that maps to audit control requirements and shrinks the storage footprint of PHI.

How does this help with HIPAA 164.314(a)(2) business associate agreements?

HIPAA 164.314(a)(2) requires that PHI sharing with business associates is limited to what is permitted by the BAA. HoundDog.ai detects every third-party SDK and API call in code, traces which PHI elements flow into each one, and lets you define a per-vendor allowlist that fails the pull request when an unapproved PHI element is added.

Does HoundDog.ai need access to production or to actual PHI?

No. HoundDog.ai runs entirely on source code in your development environment or CI pipeline. It never needs production access, never sees actual PHI, and never sends source code off your machine.

Does HoundDog.ai detect PHI flowing into LLMs like OpenAI or Anthropic?

Yes. HoundDog.ai recognizes direct and indirect AI SDKs including OpenAI, Anthropic, LangChain, and LlamaIndex, and traces which PHI fields end up inside prompts or model inputs before the call ever executes in production.

Make HIPAA Compliance a Property of Your Code

Trace every PHI flow before it ships, shrink the storage footprint of patient data, and keep your BAA inventory honest with evidence pulled directly from source.