LinkedIn Ads API 2026: AI Policies, Technical Specs, and Validation Architecture

Programmatic campaign execution on LinkedIn demands rigorous adherence to community policies, data governance, and the directives the platform imposes on AI tools through its API.

AI Development Directives

The developer policy on AI functionality establishes two indispensable operational rules for any ad generation engine:

Absolute Prohibition on Model Training: Corporate or member data obtained through the APIs (posts, comments, reactions, profiles) cannot be used to train AI models, whether proprietary or third-party. This data can only be used as temporary runtime inputs for authorized tasks such as content categorization or sentiment analysis, guaranteeing non-retention of the information.

Human-in-the-Loop Requirement: Automated publication of AI-generated content on LinkedIn without active user review is prohibited. Tools must provide an editing interface for the administrator to review, modify, and approve any synthetically generated text or image before final distribution.

Rest.li Protocol and Versioning

Requests must be directed to https://api.linkedin.com/rest/ using Rest.li protocol version 2.0.0. Each version carries a support window of 12 months from release in the YYYYMM format.

Authorization: Bearer {Access_Token}
X-Restli-Protocol-Version: 2.0.0
Linkedin-Version: 202601
Content-Type: application/json

Permission and Role Matrix

Capability	OAuth Scopes	Required Role	Endpoints
Campaign Management	rw_ads, r_ads	CAMPAIGN_MANAGER	/rest/adCampaigns
Analytics	r_ads	VIEWER	/rest/adAnalytics
Lead Sync	r_marketing_leadgen_automation, rw_ads	CAMPAIGN_MANAGER	/rest/leadForms
Conversions API	rw_conversions, r_ads	CAMPAIGN_MANAGER	/rest/conversions
Audiences (DMP)	rw_ads	CAMPAIGN_MANAGER	/rest/dmpSegments

Critical Deprecations

The legacy endpoints /adForms and /adFormResponses have been retired. Lead retrieval operations must use /leadForms and /leadFormResponses.

In /adAnalytics, declaring fields via fields= is mandatory, with a maximum limit of 20 fields per query. Requests exceeding the limit return HTTP 400.

Technical Specifications by Format

Format	Dimensions	Intro Text	Headline
Single Image	1200x628 (1.91:1), 1200x1200 (1:1), 720x900 (4:5)	150/600 chars	70/200 chars
Video	16:9, 1:1, 4:5, 9:16	150/600 chars	70/200 chars
Carousel	1080x1080 per card	255 chars	45 chars per card
Document	PDF/DOC/PPT up to 300 pages	150/600 chars	70/200 chars
Lead Gen Form	Integrated	160 chars (detail)	60 chars (headline)
Conversation	Banner 300x250 (desktop)	8,000 chars	60 chars

Four-Tier Validation System

For an ad generation engine, rules divide into four severity levels:

Tier 1, Official API Requirement (Hard Constraint): Protocol validation, character limits, maximum analytics fields, and mandatory fields. Halts execution with codes like ERR_TEXT_OVERFLOW or ERR_PROJECTION_LIMIT.

Tier 2, Data Governance (AI Policy): Guarantee that data sent to AI models is used only as volatile context without retention (ERR_AI_DATA_LEAK). Assign DRAFT status to all synthetically generated assets until human approval (WARN_PENDING_APPROVAL).

Tier 3, Official Recommendation: Suggest 1:1 or 4:5 formats over horizontal for mobile (REC_ASPECT_RATIO). Keep intro text under 150 characters (REC_COPY_LENGTH). Limit Lead Gen Forms to 3-4 fields (REC_FORM_FRICTION).

Tier 4, Industry Best Practice: Generate 3-5 differentiated creative variants (BP_CREATIVE_FATIGUE). Combine static and video formats with Thought Leader Ads (BP_FUNNEL_GAP).

Bidding Strategies

LinkedIn operates under a modified second-price auction model. Adjudication is defined by the Ad Score indicator, combining Bid Value with the Relevancy Score based on expected click-through rate, engagement level, landing page quality, and dwell time.

Strategy	Recommended Use
Maximum Delivery	Launch phase (first 14 days)
Cost Cap	BOFU campaigns with strict CPL targets
Manual Bidding	ABM strategies targeting specific buying committees

Sources

1. Developer AI Policy, Microsoft Learn
2. LinkedIn Marketing API Versioning, Microsoft Learn
3. Recent Marketing API Changes, Microsoft Learn
4. Conversions API, Microsoft Learn
5. DMP Segments, Microsoft Learn
6. LinkedIn Ads Policy 2026, AuditSocials
7. Single Image Ads Specs, LinkedIn
8. Video Ads Specifications, LinkedIn
9. Document Ads Specs, LinkedIn