欧洲议会最近通过《人工智能法案》，这是一个管理欧盟内人工智能使用的综合监管框架。这项立法可能会对马来西亚的所有企业，甚至广泛经济产生深远影响。
本文尝试概述人工智能法案可能影响马来西亚企业的几个关键点：
禁止的人工智能系统：该法案禁止人工智能用于某些用途，包括：
●操纵：可能影响人们行为，特别是弱势群体行为的技术。
●社会评分：根据社会或经济因素对个人进行分类。
●生物特征识别：使用脸部辨识进行大规模监控。
虽然这些禁令不太可能直接影响马来西亚企业，但这为负责任的人工智能开发建立了全球标准。
高风险人工智能系统：法案对以下领域使用“高风险”人工智能实施更严格监管：
●关键基础设施：交通和能源等领域使用的人工智能必须经过严格的安全检查。
●教育与就业：考试评分或招募中使用的人工智能应是公正和透明的。
●基本服务：用于信用评分或执法的人工智能必须尊重基本权利。
生成式人工智能（如 ChatGPT）：虽然不属于高风险，但生成式人工智能必须：
●公开人工智能产生的内容：用户在与人工智能创建内容互动时应有所警觉。
●防止非法内容：人工智能系统的设计应避免产生有害或误导内容。
●尊重版权：用于人工智能开发的训练资料必须遵守版权法。
时间表和执行：
●正式生效：预计2024年5月/6月生效。
●分阶段实施：
○禁用遭禁的人工智能-6个月。
○一般人工智能规则-1年。
○ 法案全面执行-2年。
●违规罚款：最高3500万欧元或全球年营业额的7%罚金。
欧盟人工智能法案可能对马来西亚企业和更广泛的经济产生的重大影响。
1.增加合规成本：在欧盟营运或使用欧盟开发的人工智能系统的马来西亚企业，可能会因需要适应新法规而面临成本增加，例如生成式人工智能透明度要求或高风险人工智能系统的生命周期评估。
2.市场机会：相反，《人工智能法案》也可能带来市场机遇。如果开发人工智能系统的马来西亚公司能够证明遵守人工智能法案，他们可以在欧盟找到新的市场。提供合规人工智能服务的公司也可能看到需求的增加。
3.创新和投资：人工智能法案可以刺激马来西亚人工智能领域的创新和投资。《人工智能法案》提供的明确监管框架可以增强投资者的信心，从而有可能增加对马来西亚人工智能新创公司和专案的投资。
4.资料隐私和保护：《人工智能法案》对资料隐私和保护的强调可能会影响马来西亚在这方面的法规，这可能导致马来西亚制定更强有力的资料隐私和保护法。这将使消费者受益，并有可能带来更值得信赖的数码经济。与欧盟伙伴合作的马来西亚中小企业可能面临更严格的资料管理要求。
总之，虽然欧盟人工智能法案为马来西亚企业和经济带来了挑战，但其也提供了机会。透过了解和适应这些新法规，马来西亚企业不仅可以确保合规性，还可以在全球人工智能市场上获得竞争优势。像往常一样，成功的关键是灵活性、创新和拥抱变革的意愿。
陈奕强《欧盟人工智能法对大马企业之影响》英文版：
The European Parliament has recently approved the Artificial Intelligence (AI) Act, a comprehensive regulatory framework that governs the use of AI within the European Union (EU). This legislation could have profound implications for all businesses and the wider economy in Malaysia.
The AI Act outlines several key points that could affect Malaysian businesses:
Prohibited AI Systems: The Act prohibits certain uses of AI, including:
●Manipulation: Technologies that could influence people’s behavior, particularly vulnerable groups.
●Social Scoring: The classification of individuals based on social or economic factors.
●Biometric Identification: The use of facial recognition for mass surveillance.
While these prohibitions are unlikely to directly affect Malaysian businesses, they establish a global standard for responsible AI development.
High-Risk AI Systems: The Act enforces stricter regulations on “high-risk” AI used in:
●Critical Infrastructure: AI used in sectors like transport and energy must undergo rigorous safety checks.
●Education and Employment: AI used in exam scoring or recruitment should be unbiased and transparent.
●Essential Services: AI used in credit scoring or law enforcement must respect fundamental rights.
Generative AI (like ChatGPT): While not classified as high-risk, generative AI must:
●Disclose AI-Generated Content: Users should be aware when they’re interacting with AI-created content.
●Prevent Illegal Content: AI systems should be designed to avoid generating harmful or misleading content.
●Respect Copyright: Training data used for AI development must comply with copyright laws.
Timeline and Enforcement:
●Official Law: Expected to be in effect by May/June 2024.
●Phased Implementation:
○Ban on prohibited AI - 6 months.
○General AI rules - 1 year.
○Full Act enforcement - 2 years.
●Non-Compliance Fines: Up to €35 million or 7% of global annual turnover.
The EU AI Act could have significant implications for Malaysian businesses and the wider economy.
1.Increased Compliance Costs: Malaysian businesses operating in the EU or using EU-developed AI systems may face increased compliance costs due to the need to adapt to new regulations, such as transparency requirements for generative AI or the lifecycle assessment of high-risk AI systems.
2.Market Opportunities: Conversely, the AI Act could also present market opportunities. Malaysian companies that develop AI systems could find a new market in the EU if they can demonstrate compliance with the AI Act. Companies offering AI compliance services could also see increased demand.
3.Innovation and Investment: The AI Act could stimulate innovation and investment in Malaysia’s AI sector. The clear regulatory framework provided by the AI Act could boost investor confidence, potentially leading to increased investment in Malaysian AI startups and projects.
4.Data Privacy and Protection: The AI Act’s emphasis on data privacy and protection could influence Malaysia’s own regulations in this area, potentially leading to stronger data privacy and protection laws in Malaysia. This could benefit consumers and potentially lead to a more trusted digital economy. Malaysian SMEs working with EU partners might face stricter data governance requirements.
In conclusion, while the EU AI Act presents challenges for Malaysian businesses and the economy, it also offers opportunities. By understanding and adapting to these new regulations, Malaysian businesses can not only ensure compliance but also potentially gain a competitive advantage in the global AI market. As always, the key to success will be flexibility, innovation, and a willingness to embrace change.
