Adapting the 'Safe Harbour' Rules for Generative AI Outputs — Centred on the Hierarchical Obligations of 'Notice and Necessary Measures'

Copyright infringement arising from generative AI outputs can be categorised into two types: prompt-induced generation and model-memory reproduction. This typological differentiation has caused a structural shift in the prerequisites for the application of traditional notice rules, manifesting specifically in the loss of structural uniformity among the three parties involved, the dissolution of the locatability of infringing content, and the gradation of service providers' capacity to address such issues. The 'notice' rule and the 'knowledge' rule can be functionally distinguished: the former serves a procedural response function, whilst the latter serves a substantive function of determining fault. Under this regulatory framework, a tiered system of obligations should be established, grounded in the quality of the notice, bounded by technical capabilities, and operating through a mechanism of dynamic matching. Defective notices should only trigger obligations to forward the notice and rectify it, whilst valid notices should correspond to obligations to filter keywords and prompts; enhanced notices may extend to obligations to block outputs at the model level. The boundaries of obligations between application-layer developers and foundational model developers should also be distinguished on a binary basis, ensuring that disposal obligations are aligned with technical feasibility.

Authors

Institutions

Publication Details

Journal
Lecture Notes in Education Psychology and Public Media
Published
2026-11-17
DOI
https://doi.org/10.54254/2753-7048/2026.jo39103
Primary Topic
Law, AI, and Intellectual Property
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Adapting the 'Safe Harbour' Rules for Generative AI Outputs — Centred on the Hierarchical Obligations of 'Notice and Necessary Measures'

Zirui Lu
1 links
Lecture Notes in Education Psychology and Public Media
Law, AI, and Intellectual Property
article

Adapting the 'Safe Harbour' Rules for Generative AI Outputs — Centred on the Hierarchical Obligations of 'Notice and Necessary Measures'

Zirui Lu
article en

Abstract

Copyright infringement arising from generative AI outputs can be categorised into two types: prompt-induced generation and model-memory reproduction. This typological differentiation has caused a structural shift in the prerequisites for the application of traditional notice rules, manifesting specifically in the loss of structural uniformity among the three parties involved, the dissolution of the locatability of infringing content, and the gradation of service providers' capacity to address such issues. The 'notice' rule and the 'knowledge' rule can be functionally distinguished: the former serves a procedural response function, whilst the latter serves a substantive function of determining fault. Under this regulatory framework, a tiered system of obligations should be established, grounded in the quality of the notice, bounded by technical capabilities, and operating through a mechanism of dynamic matching. Defective notices should only trigger obligations to forward the notice and rectify it, whilst valid notices should correspond to obligations to filter keywords and prompts; enhanced notices may extend to obligations to block outputs at the model level. The boundaries of obligations between application-layer developers and foundational model developers should also be distinguished on a binary basis, ensuring that disposal obligations are aligned with technical feasibility.

Lecture Notes in Education Psychology and Public MediaVol. 154(1)
Yunnan University (CN)
Peace, justice, and strong institutions
Law, AI, and Intellectual Property
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon