The regulation of AI is one of the most significant challenges of our time. As countless organisations and individuals harness the technology’s potential, the rapid acceleration of AI has led to extensive calls to ensure its use is ethical and trustworthy.
With the passing of the EU’s AI Act in March 2024, adhering to regulations has become increasingly paramount. Representing the first-ever legal framework on AI, the AI Act aims to address the risks of AI and position Europe as a leading global player, with the US, UK and China now working to publish their own guidelines.
Since 1901, BSI has been creating an enduring legacy of transformation for the benefit of all. Today, the organisation is well established to tackle today’s challenges, including ensuring the responsible use of AI for the benefit of society.
Operating within the Testing Inspection Certification (TIC) industry, BSI is one of the world’s largest certification bodies, working to ensure businesses adopt best practices mandated by regulations or voluntarily embraced for improved efficiency and service quality.
With a legacy spanning 120 years, BSI's commitment to ethical conduct and excellence is upheld through its Royal Charter status, signifying a dedication to societal impact beyond profit motives.
“What we aim to achieve is to make a meaningful impact towards fostering a fair society and a more sustainable world,” explains Manuela Gazzard, Group Director of Regulatory Services at BSI. “This involves tackling cutting-edge problems and advancing societal progress.”
Tackling the task of AI regulation is one of those cutting-edge challenges. Today, BSI's foray into AI regulation marks a pivotal moment in its history. Mark Thirlwell, Managing Director of BSI’s AI Regulatory Services Business Unit, outlines the need to ensure AI systems meet stringent standards.
“We're not here to stifle innovation or impede AI's progress,” he says. “We aim to facilitate innovation. For AI to reach its full potential, it must adhere to principles of fairness, accessibility and transparency.”
BSI is actively assisting organisations in bringing AI products to market in a manner that is fair and transparent – with an initial focus on medical applications.
“Currently, our focus is primarily on medical devices, although we anticipate extending our services to all industries where AI is utilised in high-risk scenarios. We already have more than 40 clients in the medical field whom we are reviewing against ‘state-of-the-art’ practices,” Thirlwell adds.
"Having worked across various industries, I've never felt more fulfilled in my role. Witnessing the impact of our clients' work – whether it's aiding children with ADHD, detecting cancers or identifying fetal anomalies – is truly remarkable.”
BSI helps customers navigate the AI landscape
By offering a suite of services encompassing AI management, product design assessment and algorithm auditing, BSI aims to support organisations in navigating the complex regulatory landscape and instilling trust among stakeholders.
As BSI prepares to establish itself as an AI-notified body – an organisation designated to assess the conformity of AI products before being placed on the market – Mark explains that its aim is not only to fulfil regulatory requirements but also assist the industry in enhancing safety, fairness, trustworthiness and regulatory readiness.
“We are in the process of developing capabilities and services that will fall under our role as an AI-notified body,” he says. “These services are being gradually introduced to give customers confidence in the development of their AI systems, ensuring they meet the necessary quality standards.”
Essentially, he describes that BSI’s approach can be likened to peeling layers off an onion. At the outer layer, this involves the focus on the management of AI within companies, and evaluating whether companies have effective AI management systems in place, encompassing people, processes and governance.
Here, BSI assesses compliance against the ISO/IEC 42001 standard, a standard which BSI has played a central role in developing and provides guidelines for how AI management systems should function.
“Moving inward, we assess the design of AI products on a case-by-case basis,” he explains. “This entails conducting technical design assessments to ensure that the AI's design aligns with established standards, addressing factors such as bias and accuracy.”
The innermost layer involves evaluating the actual functionality of the AI system, where a significant shift in regulatory approach occurs. “Unlike traditional medical devices, where assurance is largely derived from design and management practices, AI necessitates scrutiny of the underlying data sets and algorithms due to continuous development and reliance on training data.”
To address this, BSI has partnered with Citadel AI, a company specialising in algorithm auditing software, to offer algorithm auditing and data set testing services.
“We now provide a comprehensive suite of services to customers, particularly those operating in high-risk environments subject to the AI Act,” he adds. “This ensures they are well-prepared, transparent and compliant, enabling them to reassure investors and customers of their safety and transparency. When regulatory mandates come into effect, these companies will be in a strong position to undergo scrutiny by AI-notified bodies.”
Implications of AI regulation on customers: Navigating the transition
As Gazzard highlights, there are several potential challenges faced by customers unfamiliar with working with notified bodies, particularly startups venturing into regulated domains. The transition to AI regulation may pose hurdles, necessitating early engagement and preparedness to avoid disruptions to innovation.
“There are customers already working with notified bodies at the moment. For them, it's essentially an extension, as they are familiar with the process of working with a notified body,” she explains. “Our main concern lies with customers who have never engaged with a notified body before because their products have not previously fallen under its regulations.”
This process is rigorous – particularly for an AI market filled with startups and scaleups that may have not previously engaged with notified bodies such as BSI.
“Many startups operate in this space, which fosters innovation and introduces new concepts. However, many of these companies have never dealt with a notified body before, and accessing third-party consultancy services to prepare for this might prove challenging.”
As a result, she strongly advocates for the availability of various tools to assist, such as training courses offered by BSI and other providers. “We also produce white papers, conduct roadshows and actively engage to ensure readiness for when these regulations come into effect. Otherwise, the consequence could be a slowdown in innovation.”
BSI, being a notified body, does not offer consultancy services as this would be breach of impartiality.
Thirlwell underscores the importance of taking proactive measures, stressing the benefits of early adoption in gaining competitive advantage and fostering trust among consumers and investors. Businesses will have a two-year transition period to comply with the AI Act from it entering into force, but Thirlwell warns that action should be taken today.
“When people hear about a two-year transition period for the AI Act, it might sound like a lengthy period, but in reality, it isn't,” he says. “We can only achieve so much within that time frame. If everyone waits until just two months before the end of the transition period to apply, it will undoubtedly cause issues.
“We should be addressing the standards underlying the AI Act and the risks it aims to mitigate right now. Currently, AI is being used in various high-risk scenarios, such as medical devices, robotics in manufacturing and semi-autonomous vehicles. These are risks we must manage without delay. Let's not wait for regulation to force us into action.”
BSI's future trajectory: Pioneering solutions for a sustainable world
Looking ahead, BSI remains steadfast in its commitment to addressing society's most pressing needs. Gazzard envisions a future centred on sustainability and digitisation, with BSI playing a pivotal role in setting standards and providing innovative solutions.
As she highlights, BSI’s focus is twofold. “Firstly, there's a significant emphasis on sustainability: encompassing areas such as reducing greenhouse gases, improving waste management, establishing ethical supply chains and developing sustainable infrastructure. Secondly, we're prioritising digitisation, particularly in terms of digital governance, risk management, data stewardship and the integration of AI.
“We recognise the importance of practising what we preach,” she adds. “We must undergo our own digital transformation, and adapt our service delivery methods accordingly. We've moved beyond relying on clipboards, for instance, but there is still progress to be made.”
Gazzard also foresees a significant shift in the TIC industry towards data collection. “Rather than simply identifying issues, we'll be gathering valuable data to anticipate and prevent problems. This approach not only enhances safety but also recognises the immense value of data. Therefore, we must embark on an internal transition journey while simultaneously supporting our clients externally.”
Thirlwell echoes this sentiment, highlighting AI's potential to drive productivity and solve complex challenges such as sustainability. By leveraging its expertise and international partnerships, BSI aims to navigate the complexities of AI regulation and usher in a future guided by fairness, trustworthiness and technological advancement.
On the AI front, Thirwell explains that there is still plenty of work to be done. “Given BSI's international reputation and partnerships, I believe we can play a significant role in navigating complexities surrounding international regulations. It's simply not acceptable for AI developers to have to comply with 15 different sets of regulations from various international regulators.
“AI is a powerful force for good. Every day, we hear about its potential to enhance productivity. Humanity finds itself facing significant challenges, and I see AI as a tool to help us overcome them.”
**************
Make sure you check out the latest edition of AI Magazine and also sign up to our global conference series - Tech & AI LIVE 2024
**************
AI Magazine is a BizClik brand
