IP OwnershIP Of Proprietary AI Models Used In Autonomous Port-Logistics Optimisation.
1. Core IP Components in Port-Logistics AI Systems
A proprietary AI system used in port logistics typically includes:
Source code & algorithms → Copyright + Trade Secret
Model architecture & training techniques → Patent (if novel)
Training datasets (shipping data, AIS signals, port traffic) → Database rights / copyright
Outputs (optimized routing decisions) → Usually not protected unless human input exists
Legal Insight
Courts consistently emphasize that AI itself cannot be an owner, and ownership vests in human creators, companies, or operators.
2. Ownership Models in Autonomous Port AI
(A) Developer Ownership Model
AI company builds proprietary model → retains IP
Port authority gets license (SaaS / enterprise deployment)
(B) Joint Ownership Model
Port authority provides data + funding
Developer builds model
Ownership governed by contract (critical in practice)
(C) Employer Ownership Model
If built in-house → employer owns IP (work-for-hire principle)
3. Key Legal Issues
(i) Ownership of the AI Model Itself
Protected as software (copyright) and possibly patent
Functional ideas NOT protected—only expression
(ii) Ownership of Training Data
Shipping/logistics data often proprietary
Misuse may trigger database rights + copyright infringement
(iii) Ownership of Outputs
If fully automated → often no copyright
If human-supervised → human may claim authorship
4. Important Case Laws (Detailed Analysis)
4.1 Navitaire Inc v EasyJet Airline Co.
Facts
Navitaire developed airline reservation software. EasyJet created a similar system without copying source code.
Issue
Whether functionality of software can be protected.
Judgment
Copyright protects source code only, not functionality or ideas.
Relevance to Port AI
Logistics optimisation logic (e.g., berth scheduling algorithms)
❌ NOT protected
Actual implementation code
✅ Protected
👉 Competitors can build similar port AI systems without infringement if no copying occurs.
4.2 Whelan v Jaslow
Facts
Defendant created software with similar structure and purpose.
Issue
Whether non-literal elements (structure, sequence, organisation) are protected.
Judgment
Extended protection beyond code to program structure.
Relevance
AI architecture in logistics systems (e.g., pipeline design, modules)
may be protected if substantially similar.
👉 Stronger protection compared to Navitaire (jurisdiction matters).
4.3 Thaler v Perlmutter
Facts
Stephen Thaler tried to register copyright for AI-generated artwork naming AI as author.
Judgment
Copyright requires human authorship
AI cannot be an author
Relevance
Fully autonomous logistics decisions (AI-generated outputs):
❌ No ownership unless human involvement
👉 Important for port authorities relying on automated decision engines.
4.4 Li v Liu
Facts
AI-generated image created using prompts and parameter tuning.
Judgment
Human involvement (prompting, selection, refinement) = copyrightable work
Relevance
If port operators:
tune AI parameters
select outputs
→ They may claim ownership of optimization results.
4.5 Robertson v Thomson Corp
Facts
Freelance articles used in digital database without proper rights.
Judgment
Database use must respect original context and rights
Relevance
Port AI systems rely heavily on:
vessel data
cargo logs
shipping databases
👉 Unauthorized aggregation into AI training datasets may infringe rights.
4.6 Bartz v Anthropic
Key Principle
AI training on lawfully acquired data = fair use (transformative)
But pirated data is not allowed
Relevance
Port AI developers must:
Use licensed maritime datasets
Avoid scraped proprietary logistics data
4.7 GEMA v OpenAI
Judgment
AI reproducing copyrighted content = infringement
Relevance
If logistics AI reproduces proprietary datasets or reports → liability arises
5. Patent Law Dimension
AI-based port optimisation systems may qualify for patents if they:
Provide technical solution (e.g., real-time berth allocation system)
Show novelty + inventive step
However:
AI itself cannot be inventor (globally rejected principle linked to Thaler case)
👉 Inventorship must be assigned to human developers or engineers
6. Trade Secret Protection (Most Critical in Practice)
Many port AI companies rely on trade secrets instead of patents:
Model weights
Training pipelines
Data preprocessing techniques
Why?
Avoid disclosure required in patents
Strong protection if secrecy maintained
7. Contractual Control (Dominant in Port Sector)
In real-world port logistics:
IP ownership is usually determined by:
Licensing agreements
Public-private partnership contracts
Data-sharing agreements
👉 Contracts override default IP rules in most deployments.
8. Key Legal Principles (Synthesis)
From the above cases and doctrines:
(1) Human-Centric Ownership Rule
AI cannot own IP
Ownership → developers / operators
(2) Idea–Expression Dichotomy
Functional logistics optimization logic ≠ protected
Code + structure = protected
(3) Data Legitimacy Rule
Lawful data → allowed (fair use possible)
Unauthorised data → infringement
(4) Hybrid Ownership Reality
AI systems = multi-layered ownership:
Code → developer
Data → port authority / third parties
Outputs → conditional (depends on human input)
9. Application to Autonomous Port Logistics
Example Scenario
A smart port uses AI for vessel scheduling:
| Component | Owner |
|---|---|
| AI model | Developer company |
| Training data | Port authority / shipping firms |
| Outputs | Possibly public domain OR operator-owned |
| Improvements | Depends on contract |
10. Conclusion
Ownership of proprietary AI models in autonomous port-logistics optimisation is fragmented and layered, governed by:
Copyright law (software, datasets)
Patent law (technical innovations)
Trade secrets (models and pipelines)
Contracts (most निर्णायक factor)
Final Legal Position
No single “owner” of the entire AI ecosystem
Instead, a bundle of rights distributed across stakeholders
Courts increasingly focus on:
human contribution
data legality
market impact of AI outputs
RELATED Blog
comments