Add AI-Advertising Security guides - #1319
Open
elang2 wants to merge 1 commit into
Open
Conversation
Adds two full-length control-family guides to `pages/controls/`, companions to the OWASP CheatSheetSeries "AI-Powered Advertising Systems Security Cheat Sheet" (OWASP/CheatSheetSeries#2380, addresses OWASP/CheatSheetSeries#2323). The cheat sheet is a concise summary of these longer references, per the split @jmanico proposed in the review thread. Two guides: - AI-Powered Advertising Systems Security. Classical machine learning on the live ad-serving path across programmatic RTB, Programmatic Guaranteed, PMP, direct-sold, search, retail media, CTV/OTT, DOOH, native, and mobile mediation. Covers consent as a model-invocation gate (GDPR Art. 4(4), DSA Art. 26(3), EU AI Act Annex III point 4(a), CJEU C-604/22 on TC strings, COPPA), training-corpus admission (NIST AI 100-2e2025 poisoning taxonomy, backdoor scanning), inference-time defenses (extraction, membership inference, model inversion, test-time evasion, sponge attacks), outcome-event and API-boundary integrity (RFC 9421 HTTP Message Signatures, ads.cert 2.0), and AI supply chain (OpenSSF Model Signing, CycloneDX ML-BOM, ads.txt / SupplyChain object provenance). - LLM and Generative AI Security in Advertising. Companion covering what changes when LLMs, VLMs, and generative endpoints enter the serving path. Covers the LLM input boundary (prompt injection at publisher HTML, landing pages, brief fields, submitted images and audio, on-device LLMs in cross-origin ad iframes), retrieval trust and the fine-tune supply chain, cross-tenant isolation and system- prompt leakage, agentic ad-management tool authority, and generative- creative provenance under EU AI Act Art. 50(2) and 50(4) (C2PA 2.1 hard bindings, per-rendition re-signing). Both guides follow the threat, defense, evidence, and runtime signal structure @planetlevel suggested on the proposal thread.
Contributor
|
If they're meant to be cheat sheet then add them to the cheat sheets project? |
Author
|
Thanks. The concise cheat sheet lives at OWASP/CheatSheetSeries#2380. These are the longer companion references on the wiki, per the split @jmanico proposed there (OWASP/CheatSheetSeries#2380 (comment)). Is |
Contributor
|
I'll try to catch up on this Tuesday. |
Contributor
There was a problem hiding this comment.
🟢 Approval recommended
The changes are additive documentation pages, and the only issues found are minor navigability/title-consistency fixes (internal linking) that do not affect correctness.
Pull request overview
This PR adds two new OWASP Controls pages that act as companion guides for securing AI-powered advertising systems—one focusing on classical ML controls across the ad-serving path and one focusing on LLM/generative-AI controls (prompt injection, RAG trust, multi-tenant isolation, agentic tooling, and provenance).
Changes:
- Add a classical-ML-focused “AI-Powered Advertising Systems Security” guide covering consent gating, corpus protection, inference hardening, outcome-event/API integrity, and AI supply chain controls.
- Add an “LLM and Generative AI Security in Advertising” guide covering LLM input boundaries, retrieval/fine-tune supply chain, cross-tenant isolation & prompt leakage, agentic campaign management, and generative provenance/runtime controls.
- Introduce extensive reference sections (standards, academic anchors, regulatory mapping) to support the guidance.
File summaries
| File | Description |
|---|---|
| pages/controls/AI_Powered_Advertising_Systems_Security.md | New classical-ML advertising security companion guide with structured controls and references. |
| pages/controls/LLM_Generative_AI_Advertising_Security.md | New LLM/generative advertising security companion guide covering injection/RAG/agents/provenance and references. |
Review details
Suppressed comments (3)
pages/controls/AI_Powered_Advertising_Systems_Security.md:276
- This companion-sheet mention is unlinked and uses a different title than the companion page, which reduces discoverability.
Creative reaches three AI consumers regardless of mode: LLM/classifier review (see the companion *LLM & Generative AI Security in Advertising Cheat Sheet*), the renderer (defer to the [CSP](https://cheatsheetseries.owasp.org/cheatsheets/Content_Security_Policy_Cheat_Sheet.html) and [XXE Prevention](https://cheatsheetseries.owasp.org/cheatsheets/XML_External_Entity_Prevention_Cheat_Sheet.html) cheat sheets), and the provenance chain (also the companion sheet). For the classifier-hygiene rules that belong in this sheet:
pages/controls/AI_Powered_Advertising_Systems_Security.md:340
- The related-cheat-sheet bullet refers to the companion sheet without a link and with a title that doesn’t match the companion page title/front-matter.
- *LLM & Generative AI Security in Advertising Cheat Sheet*: proposed companion sheet covering prompt injection, RAG, generative endpoints, agentic bidders, and deepfake defenses
pages/controls/LLM_Generative_AI_Advertising_Security.md:302
- In the related-cheat-sheets list, Sheet 1 is referenced without a link and with a title that doesn’t match the Sheet 1 page title/front-matter.
- *AI-Powered Advertising Systems Security Cheat Sheet*: Sheet 1 of this series (addresses OWASP CheatSheetSeries issue #2323)
- Files reviewed: 2/2 changed files
- Comments generated: 2
- Review effort level: Lite
💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.
|
|
||
| ## Introduction | ||
|
|
||
| Modern advertising platforms make AI-driven financial decisions on every ad request: what to bid, which product to rank first, whether a page is brand-safe, whether traffic is real, which touchpoint gets credit for a conversion. This cheat sheet gives builders the specific controls that stop the AI itself from becoming the attack surface. It applies across every ad-serving mode: programmatic, search, retail media, CTV/OTT (Connected TV and over-the-top streaming ads), DOOH (Digital Out-of-Home: billboards and digital screens), direct-sold, native, and mobile mediation. It covers classical machine learning (ML) on the live serving path (bid, floor, rank, recommendation, quality-score, attribution, brand-safety, invalid-traffic (IVT) / fraud detection). Controls specific to large language models (LLMs) and generative AI are in the companion sheet, *LLM & Generative AI Security in Advertising Cheat Sheet*. |
|
|
||
| ## Introduction | ||
|
|
||
| Modern advertising platforms use LLMs (large language models), VLMs (vision-language models), and generative endpoints on the live serving path for publisher-page contextual review, ad-creative policy review, autonomous campaign management, generative ad copy, and multimodal deep-fake detection. This is Sheet 2 of the OWASP AI-Advertising Security series; read Sheet 1, the *AI-Powered Advertising Systems Security Cheat Sheet*, first for the ad-serving primer (buyer/DSP, seller/SSP, publisher, auction, outcome-event boundary) and the classical-ML controls that gate consent, protect training corpora, defend inference, and secure outcome events. The five numbered sections here follow the same threat, defense, evidence, and runtime structure. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Two guides that serve as companions to the AI-Advertising Security cheat sheet.
The first covers classical machine learning security controls on the ad-serving path across all ad-serving modes.
The second covers LLM and generative-AI security controls, including prompt injection, retrieval trust, cross-tenant isolation, agentic ad management, and generative provenance.