AI GUIDE FOR BOARDS IN SINGAPORE
Section V: Organisational Resilience and Capability
11.6 Intellectual Property
AI changes how intellectual property (IP) is created, used, protected and governed.
rely on third-party generative AI platforms or external foundation models. • Copyright and licensing exposure: AI systems trained on broad or external datasets may generate outputs that resemble copyrighted or proprietary material. Boards should ensure management maintains approval workows, legal review processes and usage guardrails for customer-facing, regulated, or commercially sensitive outputs. • Employee and enterprise IP rights: Boards should ensure employment agreements, acceptable-use policies and IP clauses appropriately address AI- assisted work, ownership of AI-enhanced creations, and use of external AI tools within enterprise environments. Boards should be aware of exposure to copyright or licensing risks, especially when working with third-party vendors and external contractors.
11.6.1 IP risk landscape
Boards should recognise that AI-related risks may arise from training data, AI-generated outputs, third-party vendor models, licensing restrictions and unclear ownership rights, as shown in the AI-related IP Risk Landscape . Boards should ensure management understands what rights the organisation holds over AI-generated outputs, and what legal or contractual limitations apply. Key governance considerations include: • Ownership of AI outputs: Boards should ensure organisations clarify ownership rights over AI-generated content, commercial usage rights and any jurisdictional limitations aecting enforceability. is is particularly important where organisations
AI-Related IP Risk Landscape
DATA SOURCES: Internet Scraping, Public Datasets, Licensed Databases, Proprietary Content. LICENSING RISK: Copyright Infringement, Data Use Rights, Terms of Service Non-Compliance. 1. TRAINING DATA (Licensing Risk)
INPUT
2. INFERENCE/GENERATION (Internal / Vendor)
INTERNAL DEVELOPED Proprietary weights, algorithms, architecture
VENDOR MODEL (API) Third-party platforms e.g., OpenAI, Anthropic, Google
AI MODEL
MODEL RISK: Data Leakage, Vendor Terms, Indemnification Issues.
INFLUENCE / GENERATION
3. GENERATED OUTPUTS (Ownership Risk) OUTPUT TYPES: Content, Art, Software Code, Designs.
OWNERSHIP RISK: Lack of Copyrightability (No Human Author), Ambiguous IP Ownership (Vendor vs. User), Direct Infringement by Output (Accidental Similarity).
102
Made with FlippingBook Digital Proposal Creator