Issues of regulating Generative AI (GAI)

Source: The post issues of regulating generative AI has been created, based on the article “Digital jurisprudence in India, in an AI era” published in “The Hindu” on 3rd July 2024

UPSC Syllabus Topic: GS Paper2- Governance-Government policies and interventions for development in various sectors and issues arising out of their design and implementation.

Context: The article discusses the legal challenges of regulating Generative AI (GAI), focusing on liability for content, copyright issues, and privacy concerns. It suggests that existing laws struggle to address these issues effectively and proposes measures for better legal frameworks to manage AI’s impact.

For detailed information on Generative AI read this article here

What are the issues of regulating Generative AI (GAI)

1. Liability for Content: Determining responsibility for content created by Generative AI is complex. For example, a US radio host sued OpenAI for defamation by ChatGPT.
2. Copyright Challenges: Existing laws like India’s Copyright Act of 1957 don’t address AI-generated works. The 161st Parliamentary Standing Committee Report highlighted this gap.
3. Privacy Concerns: Generative AI can’t “unlearn” data, conflicting with rights in the Digital Personal Data Protection Act, 2023, like “right to erasure.”

What initiatives have been taken to address the issues of regulating Generative AI (GAI)

1. Legal Provisions:
2. Digital Personal Data Protection Act, 2023: Introduced “right to erasure” and “right to be forgotten,” addressing privacy concerns but raising questions about its practical application to AI, which cannot unlearn data once trained.
3. Safe Harbour Protections: Section 79 of the IT Act, as reinforced by the Shreya Singhal judgment, protects intermediaries from liability for user-generated content, provided they adhere to set guidelines. The application to AI, however, remains unclear.

For more information on Shreya Singhal judgment read this article here

1. Judicial Initiatives;
2. K.S. Puttaswamy Judgment (2017): Established a strong foundation for privacy laws, leading to the Digital Personal Data Protection Act, 2023, which addresses data privacy but does not fully resolve issues with AI data usage.

2.Judicial Interpretation: Courts, like in the case of Christian Louboutin Sas vs Nakul Bajaj and Ors (2018), have been pivotal in defining the scope of intermediary liability, impacting how Generative AI might be treated under current laws.

What should be done?

1. Sandbox Approach: A temporary immunity or sandbox approach is proposed for Generative AI platforms. This method allows platforms to operate with some legal exemptions while gathering data and experiencing real-world scenarios that can inform future regulations.
2. Data Rights and Responsibilities: The process of data acquisition for Generative AI training requires a complete overhaul. This involves ensuring legal compliance and proper compensation for the intellectual property used in training models, possibly through revenue-sharing or licensing agreements.
3. Licensing Challenges: Licensing data for Generative AI needs centralized platforms for data licensing. This would help simplify the process for developers and ensure the integrity and appropriate use of data, similar to solutions found in other industries like stock photography.

Question for practice:

Discuss the effectiveness of existing legal frameworks in regulating Generative AI and the proposed measures to enhance them.