FAQs: Navigating Legal Developments in Generative AI

What is Generative AI?

Generative AI refers to algorithms capable of generating new content, such as text, images and music  by learning patterns from existing data. Popular examples include ChatGPT, DALL-E and other large language models (“<span class="news-text_medium">LLMs</span>”).

How is Generative AI Different from Traditional AI Models?

Traditional AI focuses on classification and decision-making based on existing data, while generative AI creates new data and outputs by understanding patterns from the provided dataset.

Why are Legal Issues Arising Out Of Generative AI?

Generative AI poses new challenges around intellectual property, privacy, content liability and data ownership. Concerns also include biases, misuse and security risks due to AI’s capability in generating synthetic content which may appear authentic.

What Legal Actions Have Been Taken In Relation To Generative AI Systems?

Recent lawsuits have targeted organisations for alleged copyright infringement due to the unauthorised use of copyrighted works in training datasets. There have also been privacy-related claims over misuse of personal data and the inappropriate deployment of generative AI tools.

How Are Regulators Responding to Generative AI?

Regulators globally are increasingly focused on establishing frameworks for AI accountability and transparency. For example:

• The EU AI Act proposes strict measures for high-risk AI systems
• The UK is taking a sectoral approach to AI regulation with a white paper outlining guidelines for different industries
• The US has emphasised voluntary guidelines with recent legislative developments at state levels on AI governance

What Are the Ethical Considerations Around Generative AI?

Key ethical issues include:

• The potential for AI-generated misinformation or deepfakes
• The risk of reinforcing biases and discrimination present in datasets
• The lack of accountability for the consequences of AI-generated content

How Are Businesses Expected to Comply With Emerging AI Regulations?

Businesses should be aware of requirements such as:

• Demonstrating transparency in AI models and disclosing the use of generative AI tools
• Implementing robust data protection policies to secure training data and model outputs
• Conducting AI impact assessments to mitigate risks related to discrimination, security and ethical concerns

What Are the Intellectual Property (“IP”) Implications of Generative AI?

Generative AI’s outputs raise questions about ownership. If AI-generated works are based on training data owned by third parties, issues of copyright infringement and originality come to the forefront. Recent legal cases explore whether AI-generated content can be copyrighted and who holds the rights in relation to it.

How Can Companies Protect Their IP When Using Generative AI?

To safeguard their IP, businesses should:

• Review the terms of use for generative AI models and third-party datasets
• Implement internal IP policies to monitor AI-generated content and its sources
• Secure appropriate licences for training data and validate AI output for potential IP conflicts

What Role Does Human Oversight Play in the Deployment of Generative AI?

Human oversight remains crucial to ensuring generative AI outputs are accurate, ethical and legally compliant. Businesses are advised to implement review processes, especially in high-risk areas such as content creation, marketing and automated decision-making.

What Impact Does the EU AI Act Have on Generative AI?

The EU AI Act introduces stringent compliance obligations for high-risk AI applications, which can include generative AI systems. These obligations entail risk management measures, transparency requirements and enhanced scrutiny of the training data used.

How Can Companies Prepare for Evolving AI Regulations?

Businesses should establish cross-functional AI governance teams, engage in regular compliance audits, and stay updated on evolving regulatory standards. Legal teams should focus on risk assessments and scenario-planning based on the anticipated regulations.