Prove what data your AI systems process. Before the EU AI Act asks you to.

Build an accurate AI inventory directly from source code, trace personal and sensitive data flowing into LLM prompts and AI APIs, and produce the Article 11 technical documentation EU AI Act compliance expects. Code-based evidence, before anything reaches production.

Where the EU AI Act stands today

The regulation is in force, with prohibitions and General-Purpose AI rules already enforceable. The Digital Omnibus agreement of 7 May 2026 deferred the headline high-risk deadlines, but several obligations apply immediately and the work to meet them takes time.

€35M / 7%

Maximum fine under Article 5

Of global annual turnover, whichever is higher, for prohibited AI practices. Higher than GDPR's 4 percent ceiling.

Aug 2026 · Dec 2027

Two compliance horizons

Article 50 transparency obligations apply from 2 August 2026. Annex III high-risk obligations deferred to 2 December 2027 under the Digital Omnibus.

50%+

Of organizations lack an AI inventory

Industry readiness analysis through 2026. Without a systematic inventory, risk classification and Article 11 documentation are not possible.

Four gaps between code and EU AI Act readiness

Compliance leaders know which Articles apply. The harder question is how to prove the application behaves the way the documentation says it does.

No accurate AI inventory

Engineering ships OpenAI, Anthropic, Gemini, and agent-framework integrations through pull requests. None of these reach procurement.

AI Act Art. 6 · Art. 11

Personal data flowing into LLMs unseen

PII exposures to AI integrations are rarely intentional. They happen as codebases grow.

AI Act Art. 10 · GDPR Art. 9

Article 11 documentation is a moving target

Agile teams ship every week. Technical documentation written at design time is stale before the next sprint.

AI Act Art. 11 · Annex IV

The GDPR and AI Act overlap is dual exposure

Personal data entering an AI system is governed by both regimes. Two regulators, two penalty schedules, one root cause.

GDPR Art. 32 / 35 · AI Act Art. 10

How the EU AI Act classifies AI systems

The AI Act is risk-based. Obligations escalate with the potential harm a system could cause. The four tiers determine what your team has to produce and which deadline applies to you.

Minimal Risk

Largely unregulated

Spam filters, recommendation engines, inventory optimization, and most general AI use. Providers are encouraged to adopt voluntary codes of conduct. No specific obligations.

Limited Risk

Transparency obligations (Article 50)

Chatbots, generative AI, emotion-recognition systems, and biometric categorization. Users must be told they are interacting with AI. AI-generated content must be machine-readable as synthetic. Applies from 2 August 2026.

High Risk

Conformity assessment, documentation, oversight

Annex III systems in recruitment, credit scoring, education, employment, law enforcement, migration, and justice. Plus Annex I product-embedded AI in regulated products. Risk management, data governance, technical documentation, logging, human oversight, and registration in the EU database.

Unacceptable Risk

Banned under Article 5

Social scoring, subliminal manipulation, untargeted facial scraping, emotion recognition in workplaces and schools, real-time biometric identification in public spaces. New for December 2026: AI-generated CSAM and non-consensual intimate imagery.

Code-based evidence for EU AI Act documentation, AI inventory & privacy reviews.

At development speed. Build your AI inventory and Article 11 technical documentation from source code, 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

What HoundDog.ai produces, per Article

A side-by-side view of where the regulation makes specific requirements and the code-based evidence HoundDog.ai surfaces for each. Designed to be shared with auditors and DPOs.

Requirement What it asks for Code-based evidence from HoundDog.ai
AI inventoryArt. 6, foundational Identify which AI systems the application uses so risk tier can be assigned. Static analysis enumerates every AI integration in source code, including SDK-based and indirect agent-framework usage, with file and call-site provenance.
Data governanceAI Act Art. 10 Document categories of personal and sensitive data processed by AI systems. Taint-flow analysis traces 100+ data types into each AI sink, each tagged PII, PHI, PIFI, CHD, or SECRET, with a sensitivity level.
Risk managementAI Act Art. 9 Identify, evaluate, and mitigate risks throughout the system lifecycle. Risky data flows are surfaced at PR time with severity rating, providing a continuous risk register tied to code changes rather than periodic reviews.
Technical documentationAI Act Art. 11, Annex IV Maintain records of design, data, validation, and monitoring sufficient for authority review. Every scan produces reproducible evidence of which AI systems are called, what data they receive, and from where, exportable as audit-ready artifacts.
Record-keepingAI Act Art. 12 Automatic logging of events relevant to risk identification and post-market monitoring. Code-level provenance for every detected flow: repository, file, line, function. Persistent and queryable across scans, with diffs between runs.
TransparencyAI Act Art. 50, from Aug 2026 Disclose AI interactions to users. Mark AI-generated content as synthetic. An inventory of the AI integrations detected, and the data each receives, helps determine which Article 50 disclosures apply per feature.
Privacy by designGDPR Art. 25 Build minimization, access controls, and protection into the architecture from the start. CI integration blocks unapproved AI data sharing in pull requests, shifting privacy enforcement to design and development phases.
DPIAGDPR Art. 35 Assess and document high-risk processing before it begins. Auto-generate Data Protection Impact Assessments pre-populated with detected data flows, AI integrations, and identified risks.
Records of ProcessingGDPR Art. 30 Maintain a register of processing activities and subprocessors. New AI integrations and subprocessors appear as suggested RoPA edits for privacy team review and approval, traceable to the code change.

GDPR keeps applying in the AI era

The AI Act does not replace GDPR. Personal data flowing into AI systems is still governed by GDPR, with full penalties. The two regimes overlap at every layer of an AI feature, so the safest assumption is that both apply.

TopicWhat this means for LLM usePrimary legal basisMaximum penalties
Lawful basisA valid lawful basis must be selected before any personal data is processed by the model or provider.GDPR Art. 6Up to 20M EUR or 4% of global revenue
Special categoriesSpecial categories cannot be processed unless an Article 9 exception applies with strong safeguards.GDPR Art. 9Higher tier penalties
Privacy by designControls for minimization, access, and protection must be built into the architecture from the start.GDPR Art. 25Higher tier penalties
Security of processingEncryption, logging, and strong access controls must be in place across AI data paths.GDPR Art. 32Penalties scale with severity
Transparency and user rightsUsers must be informed and allowed access, correction, deletion, and objection over their data.GDPR Art. 12-15, 21Penalties vary
International transfersValid transfer mechanisms and documented assessments are required when AI providers operate outside the EU.GDPR Art. 44-49Penalties vary
Prohibited AI practicesSocial scoring, manipulation, untargeted biometric scraping, and emotion recognition at work or school are banned.AI Act Art. 5Up to 35M EUR or 7% of global turnover

Typical questions come from different teams

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

An AI governance team asks which AI integrations exist and what sensitive data reaches them.

Which AI services receive sensitive data in our code?

HoundDog.ai Acme Corp
New chat Which AI services receive sensitive data in our code? ···
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 AI services receive sensitive data in our code?
Searching…
Searched the datamap
Opened a page
One AI service receives sensitive data in your scanned code: OpenAI.
Data element: Medical History (PHI, high sensitivity).Where: the main branch of the hounddogai/hounddog-test-python repository, in one file.Dataflows: one confirmed dataflow, and it is risky.
No other AI data sinks have confirmed sensitive-data flows. I've opened the Data Sinks view in the Datamap, filtered to AI sinks. To triage the risky dataflow, use the Dataflows page.
Datamap · Data Sinks
Which AI services receive sensitive data in our code? 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 Data Element Medical History PHI
HoundDog.ai
OpenAI 3PAI 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.

EU AI Act compliance questions, answered

The Articles, deadlines, and roles teams ask about most when scoping their AI Act program.

Does the EU AI Act apply to a company based outside the EU?

Yes, if the AI system is placed on the EU market or its outputs affect people in the EU. The EU AI Act has extraterritorial reach similar to GDPR. A US or UK company whose AI features are used by EU customers is in scope, regardless of where the model runs.

What changed for high-risk AI systems with the Digital Omnibus agreement in May 2026?

The Digital Omnibus provisional agreement of 7 May 2026 deferred the headline compliance dates. Annex III high-risk AI obligations move from 2 August 2026 to 2 December 2027. Annex I product-embedded high-risk systems move from 2 August 2027 to 2 August 2028. Article 50(2) watermarking moves to 2 December 2026. Prohibited practices under Article 5 and General-Purpose AI obligations remain in force unchanged, and the August 2026 transparency obligations under Article 50 still apply.

Are we a provider or a deployer when we build features on top of OpenAI, Anthropic, or Gemini?

Most teams that wrap a foundation model in a product feature are deployers under the AI Act. Substantial fine-tuning, retraining, or rebranding can reclassify you as a provider, which carries heavier documentation and conformity obligations. The boundary is interpretive, so map your AI usage carefully and document the modifications your engineering team makes.

How does HoundDog.ai help with Article 11 technical documentation?

HoundDog.ai produces code-based evidence of which AI systems your application calls, which categories of personal and sensitive data flow into each one, and which file, function, and call site each integration originates from. This evidence is reproducible across every scan, traceable to the code that generated it, and is exactly the documentation Article 11 expects organizations to maintain.

What are the maximum fines under the EU AI Act?

Violations of Article 5 prohibited AI practices carry the highest tier, up to 35 million euros or 7 percent of global annual turnover, whichever is higher. Non-compliance with other obligations such as data governance, technical documentation, and transparency carries fines up to 15 million euros or 3 percent of global turnover. Supplying incorrect information to authorities carries up to 7.5 million euros or 1 percent.

Does HoundDog.ai need access to production AI systems?

No. HoundDog.ai runs in your development environment or CI pipeline and analyzes source code statically. It never needs access to your production database, runtime data, live AI traffic, or model providers. All evidence is generated before code reaches production.

Prepare for the EU AI Act with evidence, not assumptions

Build an accurate AI inventory, trace personal data into every LLM prompt, and produce the Article 11 technical documentation your auditors will ask for. From source code, before anything ships.