Ethics vs algorithms – the legal balancing act of AI development

AS THE capabilities of artificial intelligence (AI) increase at an alarming rate, Singapore has taken steps to build and govern the ecosystem by introducing a robust Model AI Governance Framework, clear guidelines in the IMDA and Aicadium Generative AI: Implications for Trust and Governance discussion paper, and reliable AI testing infrastructures such as AI Verify.

Despite these positive strides, more remains to be seen to set the stage for a cohesive and concerted approach to regulating AI worldwide.

From a legal standpoint, several issues persist.

Navigating the uncertainties of intellectual property 

One of the major concerns arising from the use of generative AI is intellectual property rights, particularly regarding copyright law. Two crucial questions emerge: First, who owns the content generated by AI? Second, can generative AI infringe intellectual property rights?

The answer to the first question is unclear. Potential copyright contenders include the AI models or systems, the artists whose artworks were used as training data, or the users who generate the prompts. ChatGPT’s terms of use state that users are assigned all rights, title and interest in and to the generated output. However, given that the output relies on Internet data sets, it begs the question: Does OpenAI hold any intellectual property rights in the first place?

The US Copyright Office has ruled out AI ownership, stating that it will not register works produced by a machine without any creative input from a human author. Consequently, users would unlikely receive copyright protection for works solely generated by AI without human intervention.

As AI technology evolves and becomes more human-like, the possibility of AI having some form of legal personality in the future cannot be denied.

There are two areas of consideration for the second question. First, could AI models infringe copyright law by collecting copyright-protected works from the Internet, or can the defence of fair use apply?

In Singapore, these four requirements must be met for the defence of fair use to succeed:

(a) The purpose and character of the use, including whether the use is of a commercial nature or is for non-profit educational purposes;

(b) The nature of the work or performance;

(c) The amount and sustainability of the portion used in relation to the whole work or performance; and

(d) The effect of the use upon the potential market for, or value of, the work or performance.

Requirement (c) poses a challenge as each generated piece of work is likely a unique permutation of the algorithm. As of now, end users are not able to tell the extent of copyrighted work that is used in creating the output. Under requirement (d), the potential for unrestricted and widespread infringement due to the accessibility of AI models to the public diminishes the likelihood of fair use applying.

The second area of consideration is whether the output of generative AI can violate copyright. The answer to this is likely no. Current copyright laws generally protect the creative expression, but not the underlying ideas or data. Even if a case for copyright infringement can be made, the identity of the defendant remains uncertain. Lawmakers must decide whether the model developer should be held liable for training AI with copyright-protected data, or if the AI system should eventually be granted a certain degree of legal personality.

Data privacy and cybersecurity

Data privacy and cybersecurity present significant concern. Where falsehoods and misinformation are perpetuated by AI, policymakers must decide who should be held liable. Allowing Internet users to decide if and how their personal information is to be collected and used by AI developers is another aspect to consider.

Given the tendency for generative AI such as ChatGPT to make mistakes and unwittingly convey misinformation, legislation must address how to handle such falsehoods. Singapore’s Protection from Online Falsehoods and Manipulation Act enables the minister to issue a corrective direction to rectify falsehoods. However, when the author of the falsehood is an AI computer, lawmakers must reconsider the recipient of the corrective direction.

AI and the law of torts

The law of torts presents the next challenge for legislative bodies worldwide. As increasingly autonomous vehicles that are powered by AI enter the market, the law on negligence and tortious liability must adapt. The current law assumes a human driver and imposes a duty of care to other road users. However, with self-driving vehicles, determining responsibility for the decisions of the AI model becomes essential. The extent to which the AI software’s decision can be attributed to the developer becomes a critical question.

In the context of the tort of harassment, Singapore’s Protection from Harassment Act penalises individuals for causing harassment, alarm and distress. Lawmakers must also consider modifying current laws to address instances where an individual becomes distressed or alarmed due to a defamatory remark generated by an AI model. Implementing measures to safeguard against the creation of toxic content can be considered.

Europe’s bold step

On Jun 14, 2023, a draft law known as the AI Act was passed by the European Parliament. The AI Act potentially mandates greater transparency from developers of generative AI models. Key points include requiring developers to disclose copyrighted data for training AI models and implementing measures to avoid generating unlawful content. The draft AI Act may also introduce a ban on the use of live facial recognition to address data privacy risks.

The AI Act is set to be passed by the end of this year, becoming the world’s first large-scale legislation on AI if successful.  

While Singapore may not be in a rush to pass an AI Act right now, continued dedication and collaboration among stakeholders are paving the way for a more secure and ethically governed AI landscape. The anticipation remains high to achieve a unified and collaborative global approach to AI regulation.

The writer is a partner at Kennedys Law