AS artificial intelligence (AI) continues to permeate various industries, including healthcare, transportation, finance, and law enforcement, governments worldwide are striving to establish frameworks that promote the responsible and ethical advancement of AI technologies.
Notably, the European Union’s
(EU) AI Act, introduced in 2021, represents one of the most thorough efforts to
regulate AI, focusing on mitigating the risks linked to its implementation
while encouraging innovation.
For nations such as Malaysia,
which are progressively investing in AI technologies, the EU AI Act serves as a
significant reference point for formulating a comprehensive AI safety
framework.
Overview of the EU AI Act
The EU AI Act marks the first
comprehensive regulatory framework for AI introduced by a major economy.
This legislation categorises AI
systems based on their risk levels, which include unacceptable, high-risk, and
low-risk classifications. Consequently, it establishes different levels of
regulatory scrutiny corresponding to each category.
A primary aim of the Act is to
ensure that high-risk AI systems adhere to rigorous standards, which encompass
transparency, accountability, and traceability.
Additionally, by delineating
explicit rules and guidelines, the Act encourages innovation in AI while
simultaneously addressing potential risks.
Furthermore, the Act seeks to
protect individuals’ rights and freedoms, including privacy and the principle
of non-discrimination while advocating for ethical practices in AI development.
It incorporates measures related
to transparency, human oversight, data governance, and accountability,
particularly for high-risk AI systems.
These aspects hold significant
relevance for Malaysia as it confronts the challenges associated with AI
advancement.
Adapting the EU AI Act for
Malaysia’s AI safety framework
Malaysia, characterised by its
rapid economic growth and increasing investments in AI, is strategically
positioned to implement a safety and governance framework modelled after the EU
AI Act.
Nonetheless, it is essential for
Malaysia to tailor these principles to fit its distinct social, political, and
economic landscape. The following outlines several strategies through which
Malaysia can utilise the EU AI Act as a reference for developing its own AI
safety framework.
The EU AI Act employs a
risk-based methodology, classifying AI systems into categories of unacceptable
risk, high risk, and low risk.
By adopting a comparable
framework, Malaysia would enable regulators to concentrate their efforts on the
AI systems that pose the greatest potential harm while simultaneously fostering
innovation in areas deemed less risky.
To initiate this process,
Malaysia could identify high-risk sectors such as healthcare, law enforcement,
and finance, subsequently establishing stringent guidelines tailored to these
industries.
In Malaysia, the application of AI within the healthcare sector is becoming increasingly prevalent, particularly in areas such as medical diagnostics, patient monitoring, and drug discovery.
While these technologies present
considerable potential, they also carry inherent risks, particularly in
instances of failure or erroneous recommendations.
By categorising healthcare AI as
high-risk, Malaysia could mandate rigorous testing protocols, ensure
transparency in decision-making processes, and implement human oversight, akin
to the regulatory framework established by the EU.
The EU AI Act underscores the
necessity of transparency and explainability for high-risk AI systems, which is
vital for enabling individuals to comprehend the rationale behind AI-generated
decisions.
This is especially pertinent in
critical sectors such as finance, healthcare, and law enforcement.
Malaysia could adopt similar
principles by obligating AI developers to furnish comprehensive documentation
detailing the underlying logic, decision-making frameworks, and data utilised
by their AI systems.
In the context of public
administration and financial services in Malaysia, it is essential for
consumers and citizens to grasp how decisions that impact them are formulated.
To enhance trust and
accountability in AI technologies, Malaysia could require AI providers to
present easily accessible information regarding their systems, including the
models, datasets, and algorithms employed in the decision-making process.
A fundamental component of the EU
AI Act is the implementation of robust data governance protocols, particularly
concerning high-risk AI systems.
The accuracy, impartiality, and
security of data utilised by AI systems are essential for ensuring their
fairness and accountability. For Malaysia, the adoption of data protection
regulations that are consistent with the EU’s General Data Protection Regulation
(GDPR) could alleviate challenges such as data privacy infringements and
algorithmic bias.
The Personal Data Protection Act
(PDPA) serves as a foundational framework for data governance in Malaysia.
By aligning with the principles
of the EU AI Act, Malaysia can enhance its data protection policies, mandating
that AI developers uphold the integrity of personal data employed in AI systems
and avert discriminatory practices.
For instance, AI algorithms
utilised in credit assessments or recruitment processes should undergo
evaluations for bias and fairness.
The EU AI Act stipulates the
necessity of human oversight for high-risk AI systems, ensuring that
AI-generated decisions can be reviewed or overridden by human operators.
Malaysia could implement similar
requirements, making it essential for AI systems, especially in critical areas
such as healthcare, law enforcement, and public administration, to incorporate
human oversight mechanisms.
In the context of AI-driven
medical diagnostics, the principles outlined in the Act could assist Malaysia
in ensuring that human medical professionals remain integral to the
decision-making process, especially in high-stakes scenarios.
Likewise, AI applications in law
enforcement, such as facial recognition or predictive policing, should be
subject to human intervention to prevent potential misuse and uphold fairness.
– Dec 31, 2024
Source: https://focusmalaysia.my/the-eu-ai-act-as-a-model-for-malaysias-ai-safety-framework-part-1/
Comments