CHILE Law and Practice Contributed by: Carolina Cabrera, LawTech
often requires local controls, internal risk assessments and documented governance rather than extensive changes to the hyperscaler’s terms. Hyperscalers also affect negotiations with integra - tors, resellers and managed service providers. These providers cannot promise more than the underlying cloud service provides, making it essential to distin - guish what is controlled by the hyperscaler, what is managed by the integrator and what remains the cus - tomer’s responsibility under the shared responsibility model. 8.4 IP Evolution Traditional IP concepts remain relevant in Chile, but they are under pressure from current technologies and are not always sufficient on their own. Copyright, industrial property, trade secrets, know-how, contrac - tual licences and confidentiality still provide the basic framework for most technology transactions. These concepts work reasonably well for software, source code, documentation, databases, brands, technical inventions, industrial designs, confidential information and reusable know-how. However, gen - erative AI, data analytics, machine learning, cloud platforms, open-source software, APIs and automated content generation create questions that classic own - ership concepts do not fully answer.
In copyright, the main tension is authorship. Chilean copyright law is structured around works of the intel - lect and the concept of an author, pointing toward human creative contribution. Current administrative criteria are consistent with that logic: fully autono - mous AI outputs may not qualify for copyright pro - tection, while AI-assisted human creations may be protectable depending on the degree of human input. In patents, the traditional framework remains relevant, but the focus is technical contribution. INAPI guidance confirms that AI-based inventions can be assessed under ordinary patent rules if they solve a technical problem through technical means. The difficulty is avoiding abstract claims, black-box descriptions or attempts to protect algorithms or code as such. Data is another area where traditional IP is imperfect. Data itself is not always protected as IP, although databases, compilations, confidential datasets, trade secrets and contractual rights may be protected. As a result, contracts increasingly regulate data access, ownership, aggregation, anonymisation, training, AI outputs, model improvements, open source, indemni - ties and permitted use.
27 CHAMBERS.COM
Powered by FlippingBook