This article is part of the AI and CSR Series, an ongoing exploration of how artificial intelligence impacts Corporate Social Responsibility. (In previous entries, we examined topics such as AI’s implications for human rights, fair labor, the environment, etc. Now we focus on consumers.) Here we delve into ISO 26000’s “Consumer Issues” — principles that emphasize fair marketing, product safety, data protection, and transparent communication — and examine how these play out in an AI-driven marketplace.
Modern consumers interact with AI everywhere, from chatbots in retail to algorithms approving loans in finance and AI diagnostics in healthtech. These innovations bring convenience but also raise pressing questions about consumer protection, transparency, and trust. Are AI-powered products safe and free from bias? Do customers know when they’re dealing with AI, and can they understand or challenge its decisions? Addressing these questions isn’t just good practice; it’s a CSR imperative. Consumer trust and loyalty hinge on feeling protected and respected by the technologies they use. Companies that align AI development with ISO 26000’s guidance on honest communication, privacy, and safety will be better positioned to foster trust and meet rising ethical expectations.
Below, we explore key facets of responsible AI use in consumer-facing products — from AI transparency and explainability, to ethical product design, data privacy, and fairness. We highlight real-world examples (especially from 2023 onward in Asia, including Singapore) that show how businesses and regulators are striving to make AI more trustworthy. This entry provides insights for consumer rights advocates, product strategists, and trust & safety professionals on building AI systems that truly put the customer first.
Transparency is a cornerstone of both CSR and trustworthy AI. In the context of consumer products, AI transparency means being open about when and how AI is used, and providing explanations for automated outcomes. This openness directly impacts customer trust. A 2023 global study found that nearly three-quarters (73%) of people are concerned about AI’s potential risks, such as privacy breaches, manipulation, and bias. However, the same research shows strong public support for principles of trustworthy AI – notably data privacy, fairness, and transparency – and people expect organizations to uphold these high standards. In other words, consumers are more willing to embrace AI if they know it’s being used responsibly and transparently.
Real-world cases illustrate the value of transparency. For example, financial services in Singapore have embraced transparency through industry guidelines. The Monetary Authority of Singapore’s FEAT principles (Fairness, Ethics, Accountability, Transparency) set out that AI-driven decisions (like credit scoring or fraud detection) should be explainable to customers and regulators. In 2023, an MAS-led consortium released the Veritas Toolkit 2.0 to help financial institutions assess AI systems against these principles. This includes methods to explain AI models’ decisions to ensure customers are treated fairly. Such transparency is critical in banking: if an algorithm declines a loan or flags a transaction, consumers deserve to know why. Providing clear reason codes or explanations (e.g. which factors impacted a credit decision) not only meets regulatory expectations but also treats the customer with respect.
Transparency is equally important in sectors like retail and online services. Lack of disclosure can erode trust – for instance, if product recommendations or dynamic prices seem arbitrary or unfair. China has taken a bold step on this front. In late 2024, Chinese regulators issued new rules for online platforms to disclose the principles behind their algorithms to users, aiming to enhance transparency and prevent consumer harm. Platforms are now expected to inform users how content feeds or prices are determined, rather than hiding behind a “black box.” This move targets issues like “information cocoons” (where opaque recommendation AIs trap users in echo chambers) and big data-based price discrimination (where loyal customers are charged more). By forcing transparency and prohibiting exploitative algorithmic practices, regulators seek to rebuild trust in digital services. Companies in any region can take note: being candid about AI use – from labeling AI-generated content to explaining personalization features – can preempt consumer mistrust and regulatory scrutiny.
Ultimately, AI transparency is about empowering the consumer. It aligns with ISO 26000’s call for accurate information, fair marketing, and honest communication with customers. Whether it’s a shopping website explaining why it recommended a product, or a health app clarifying that an initial diagnosis was AI-assisted, transparency treats users as partners in the AI experience. This openness, paired with the ability to ask questions or get human support, strengthens the consumer’s trust that the company has nothing to hide and is using AI in good faith.
While transparency is the goal, explainability is one of the key means to achieve it. Explainable AI (XAI) refers to AI systems designed so that their functioning can be understood by humans – including developers, regulators, and importantly, consumers. In practice, explainable AI is a pillar of ethical product design for AI-powered services. It ensures that when an AI influences a consumer (such as approving a loan, personalizing a news feed, or recommending a treatment), the outcome isn’t a mysterious verdict but something that can be reasoned about. This is closely tied to ISO 26000 principles around consumer safety and fair treatment, because an inexplicable AI decision can feel arbitrary or even unjust to the person on the receiving end.
Companies are increasingly recognizing that they must build AI systems with ethics and explainability by design. A notable example is Unilever, the global consumer goods company, which has been proactive in setting internal AI ethics policies. One of Unilever’s guiding principles is that “any decision that would have a significant life impact on an individual should not be fully automated and should ultimately be made by a human,” ensuring a check-and-balance for high-stakes outcomes. They also mandate accountability (employees must take ownership of AI actions – “we will never blame the system” as their policy states) and continuous monitoring of AI performance. This ethos is a model of ethical AI product design: before rolling out an AI feature, Unilever assesses its risks, biases, and effectiveness through an AI assurance process. Only when an AI tool passes muster on safety, fairness, and transparency does it get integrated into products or processes. By embedding such governance into the design stage, Unilever can confidently deploy AI in consumer-facing applications (like marketing or customer service) knowing it aligns with their CSR values and won’t betray customer trust.
In the financial sector, explainable AI is becoming a design requirement due to both ethics and regulation. Many banks in Asia are adopting AI for credit scoring and risk assessment, but they face the challenge of ensuring these complex models remain explainable and fair. One case study in 2023 involved a Singapore bank applying the FEAT fairness assessment to its AI credit model. The model initially showed that female applicants had slightly higher loan approval rates than male applicants, which raised questions of bias. However, by using multiple fairness metrics, the bank found the difference was due to legitimate factors (women applicants, on average, had better repayment odds). This nuanced analysis – enabled by explainable AI techniques – helped the bank avoid a false alarm of discrimination while still being transparent about the model’s workings. The outcome was two-fold: the AI system was adjusted and validated to ensure it did not unfairly disadvantage any group, and the bank could confidently explain its credit decisions to customers and regulators. Designing AI with such explainability not only meets fair lending laws, but also aligns with ISO 26000’s emphasis on treating consumers fairly and equitably.
Another aspect of ethical AI design is user-centric design – thinking about how an AI feature feels to a consumer. For example, in retail and e-commerce, AI chatbots and recommendation engines should be designed to be helpful without being deceptive or intrusive. An ethical design approach might include features like letting users opt out of AI recommendations, or clearly indicating “Why am I seeing this ad?” with a simple explanation. In 2024, Chinese e-commerce platforms were instructed to allow users more control and prevent algorithms from exploiting them, such as not forcing users into tailored content bubbles and prohibiting secret price hikes for certain customers. These design choices – transparency toggles, feedback options, fairness constraints – illustrate how product teams can bake CSR principles into AI interfaces. By proactively addressing privacy, fairness, and safety at the design phase, companies demonstrate respect for the consumer’s rights and autonomy, which is the essence of ethical product design.
Beyond transparency and design, fairness and responsibility in AI use are critical to consumer protection. ISO 26000’s consumer issues include the “protection of vulnerable and disadvantaged consumers” and fair treatment in marketing and service delivery. In AI terms, this translates to ensuring that AI systems do not discriminate, unintentionally or otherwise, and that they serve all segments of the population reliably. It also means avoiding uses of AI that could manipulate or mislead consumers.
Fair AI use requires vigilance against biases in algorithms. AI systems learn from data, and if that data reflects societal biases, the AI can amplify them – leading to outcomes that are unfair for certain groups (e.g. loan algorithms redlining minority neighborhoods, or facial recognition working poorly on darker skin tones). Companies must implement processes to regularly test and audit AI for bias. We’re seeing this become standard practice in sectors like finance and hiring. In fact, the fairness assessments in Singapore’s Veritas initiative mentioned earlier are a case of industry self-regulation to catch bias early. Globally, regulators are also clear that AI’s complexity is not an excuse for discrimination – for instance, U.S. agencies in 2023 jointly stated that lenders must explain credit decisions even if AI models are complex, and ensure they comply with fair lending laws. The implication for any consumer AI service (be it credit, insurance, or even retail pricing) is that it must be able to demonstrate fairness in outcomes. Techniques like algorithmic audits, bias mitigation (e.g. rebalancing training data), and using simpler interpretable models where appropriate are all part of responsible AI use.
Emerging regulations in Asia are reinforcing fair AI practices. A striking example is how China is tackling algorithmic unfairness on platforms. The new governance guidelines (2024) explicitly ban “big data discrimination,” where companies use personal data to charge different prices to different people for the same product. This was a known issue in ride-hailing and online shopping, where loyal or less tech-savvy customers might unknowingly pay more. By outlawing this practice, and requiring platforms to not use sensitive attributes (age, income, etc.) to differentiate services, the playing field is leveled for consumers. The rules also urge protections for vulnerable groups: platforms must adapt their algorithms to better serve minors and the elderly, ensuring content and services are appropriate and accessible. These measures highlight that fairness in AI is not just a technical issue, but a social one – AI should not prey on the vulnerable or exacerbate inequality. Companies in all regions should heed this direction of travel: building fairness constraints into AI (for example, testing that an AI marketing campaign doesn’t unfairly target or exclude certain demographics) is increasingly seen as a basic requirement of consumer respect.
Another aspect of responsible AI use is consumer safety and well-being. AI can pose safety risks – consider an AI medical app giving wrong advice, or a self-driving feature making a navigation error. In healthcare tech, voices are calling for cautious integration of AI. In 2025, the ECRI Institute listed “insufficient governance of AI in healthcare” as a top patient safety concern, warning that AI errors could lead to misdiagnosis or inappropriate treatment if not properly managed. The lesson extends to any consumer service: if AI is involved in decisions affecting health or safety, companies must rigorously validate the AI’s accuracy and have fail-safes. For instance, a healthtech startup deploying an AI symptom checker should ensure clinical oversight – perhaps using AI for preliminary analysis but having a doctor review critical outputs. Similarly, an AI in a car’s navigation or driver-assist system must be tested for all scenarios and regularly updated to fix issues. Ensuring product safety in the age of AI might mean extra rounds of testing, algorithmic “stress tests,” and clearly warning users about AI’s limits (e.g. “this tool is not a medical diagnosis”). These align with ISO 26000’s mandate to protect consumers’ health and safety at all costs.
No discussion of consumer issues and AI would be complete without emphasizing data protection. AI systems often hunger for data – from personal preferences to biometric info – to function effectively. Yet, using consumer data is a privilege that must be handled with care. Privacy is a core component of consumer trust: if people fear their data is being misused or sold off by an AI-powered service, that trust evaporates quickly. ISO 26000 underlines consumer data protection and privacy as a key aspect of social responsibility, and this has only grown more urgent in the AI era.
Companies deploying AI need to adopt a privacy-by-design mindset. This means building AI systems that minimize data collection, secure the data they do collect, and are transparent about data use. For example, a retail mobile app with AI personalization should clearly inform users what data it tracks (purchase history, browsing behavior) and give easy opt-outs for non-essential data collection. In practice, leading firms in Asia are moving in this direction. Singapore’s approach to AI governance explicitly intertwines data protection with AI innovation: the government’s Model AI Governance Framework and AI Verify toolkit include principles like data governance, transparency, and accountability as pre-requisites for trustworthy AI. The idea is that an AI cannot be “responsible” if it rides roughshod over privacy. Singapore’s Personal Data Protection Commission even launched an AI Governance Testing initiative to help companies audit their AI systems for compliance with privacy and ethics guidelines. This reflects a broader trend in Asia and globally – data protection laws (from Singapore’s PDPA to Europe’s GDPR to newer laws in Thailand and Indonesia) now intersect with AI deployments. Violating data privacy not only risks regulatory penalties but public backlash.
Recent events show the cost of ignoring privacy in AI. In early 2023, for instance, Italy temporarily banned a popular generative AI chatbot over privacy concerns when it was found to be collecting personal data without proper legal basis. In Thailand, the new Personal Data Protection Committee imposed multi-million baht fines in 2024 after data leaks, underscoring that authorities will enforce privacy rights. For companies using AI, these incidents are cautionary tales: earn consumer trust by safeguarding their data. Use techniques like anonymization and encryption for any personal data fed to AI models. Only retain data as long as necessary, and obtain informed consent, especially if data is used to train algorithms. Also, prepare for data breaches or AI exposures – have response plans and be ready to notify users transparently if something goes wrong. Responsible AI means not only preventing AI from making biased decisions, but also preventing AI from becoming an all-seeing surveillance tool that consumers never agreed to.
Lastly, privacy ties back to transparency: consumers should be educated about how AI uses their data. Simple privacy notices, dashboard controls for personalization, and periodic reminders of data settings can empower users. When consumers feel in control of their information, their comfort with AI technologies rises. On the flip side, any whiff of shadowy data practices can break trust permanently. Thus, treating consumer data with respect isn’t just about compliance – it’s foundational to maintaining an ethical relationship between businesses and their customers in the AI age.
Even with the best designs and intentions, AI systems can and will make mistakes or decisions that upset consumers. What happens then is a true test of a company’s commitment to responsible AI and CSR values. ISO 26000 highlights the importance of dispute resolution and redress mechanisms for consumers – essentially, if a customer is wronged or confused, the company should have a fair and accessible way to address it. In the AI context, this means providing accountability and recourse for AI-driven actions.
Accountability in AI use implies that a human authority is ultimately answerable for the AI’s behavior. Companies should avoid the trap of deflecting blame to “the algorithm” if something goes awry. As mentioned earlier, Unilever explicitly requires that there is always a responsible person for any AI decision (“there must be a Unilever owner accountable”). This kind of policy is crucial in maintaining trust – it assures consumers that the company stands behind its AI products and will take responsibility if they cause harm or inconvenience. For example, if an AI-powered e-commerce recommendation accidentally promotes inappropriate content to a child user, the company should swiftly acknowledge the lapse and fix the algorithm, rather than denying responsibility. Many forward-thinking organizations are now setting up AI ethics committees or officers to oversee such issues, which is a positive step for accountability.
Communication is another key. When an AI makes a notable decision about a consumer, clear communication can turn a potentially negative experience into a constructive one. Take the case of content moderation on social media (which often relies on AI): if a user’s post is taken down by AI for violating rules, platforms that communicate the reason (e.g. hate speech detected, with reference to the specific term or rule) and offer an appeal process tend to retain user trust better than those that issue a mysterious “Your post was removed” with no explanation. This aligns with fair communication practices advocated in ISO 26000 – being truthful, clear, and empathetic in customer interactions. In the context of AI, it may involve labeling AI interactions (“This chat agent is AI-powered”) or providing user education (such as tutorials on how an AI-based product works and its limitations). Educating consumers increases their comfort and reduces misunderstandings. For instance, a fintech app might include a note: “Our investment recommendations are generated by AI based on your profile, but you should review them or consult an advisor before making decisions.” Such frank communication treats the consumer as an informed decision-maker, not a passive data point.
Perhaps most importantly, there must be a way for consumers to challenge or seek redress for AI decisions. If a customer feels an AI-driven outcome was wrong – say, a loan denied or an account flagged unfairly – they should have easy access to a human review or an appeals process. Regulators are starting to insist on this. The new Chinese algorithm regulations require platforms to establish clear and accessible channels for appeals so users (or workers affected by algorithms) can contest decisions. This is a powerful move to restore balance: it acknowledges that algorithms aren’t infallible and that humans must have the final say. In finance, some banks already allow customers to request manual review of automated credit decisions. In e-commerce, if a price or offer was algorithmically determined, customers should be able to contact support if it seemed unfair and receive a reasonable explanation or remedy. Listening to consumer feedback about AI outcomes can also guide improvements – if many users say an AI tool is confusing or biased, that’s a signal for developers to refine the model.
In summary, accountability and recourse form the safety net of responsible AI in consumer-facing roles. By owning up to AI’s actions, communicating proactively, and giving consumers a voice, companies reinforce that technology exists to serve people, not the other way around. This human-centric approach closes the loop on the ISO 26000 consumer issues framework: it not only prevents harm but also builds a relationship of trust and respect.
As AI becomes ever more entwined with daily consumer life – from the apps we use to the services we rely on – integrating ISO 26000’s consumer protection principles into AI development is both wise and necessary. Responsible AI isn’t a box-ticking exercise; it’s about fundamentally aligning technology with human values like safety, fairness, privacy, and transparency. The examples from recent years show a clear momentum: governments and businesses in Asia and around the world are waking up to the importance of ethical AI design. Singapore’s initiatives in AI governance and finance, China’s sweeping rules for algorithmic fairness, and companies like Unilever setting internal AI ethics standards all point to a future where consumer trust is earned through action – through transparent practices, fair outcomes, and putting people first.
For organizations, embracing these principles offers a competitive advantage. Consumer trust, once lost, is hard to regain, but companies that proactively build trustworthy AI will foster deeper loyalty and brand reputation. Imagine a healthcare provider whose AI diagnostic tool is known for being transparent and rigorously validated – patients will likely choose it over a competitor’s opaque system. Or a retail platform that guarantees no AI-driven discrimination – customers will feel more comfortable and stick with it. In contrast, AI missteps (a privacy scandal, a biased algorithm exposed) can quickly lead to public backlash and regulatory penalties. Thus, aligning AI with CSR is not just ethical, it’s strategic.
This is the seventh entry in our AI and CSR Series, continuing our journey through how AI affects corporate responsibility in each of ISO 26000’s core subjects. By focusing on consumer protection, transparency, and fair AI use, we underscore that technology should enhance the customer experience without compromising rights or well-being. Companies that heed this call will not only comply with emerging laws and standards but also contribute to a marketplace where innovation and ethics go hand in hand. In doing so, they help ensure that AI truly serves humanity – delivering convenience and personalization, while upholding the trust, safety, and dignity of every consumer.
1. What is ISO 26000 and how does it relate to AI and consumer protection?
ISO 26000 is an international standard providing guidance on social responsibility. It outlines principles on issues like consumer rights, fair practices, and safety. When applied to AI, ISO 26000’s consumer protection aspect means ensuring AI-driven products are safe, advertised honestly, respect customer data privacy, and treat consumers fairly (e.g. no deceptive AI marketing or biased algorithms).
2. How does AI transparency affect consumer trust?
AI transparency significantly boosts consumer trust by demystifying how automated decisions are made. When companies openly communicate that an AI is being used and provide explanations for its outputs, consumers feel respected and confident in the service. Conversely, “black box” AI that offers no insight can breed suspicion or frustration, undermining trust.
3. What are some real-world examples of responsible AI for consumers?
One example is Singapore’s banking sector, where banks adopted the FEAT principles (Fairness, Ethics, Accountability, Transparency) to govern AI in credit scoring. This led to toolkits that ensure AI decisions on loans are explainable and fair. In retail, Chinese regulators in 2024 required e-commerce platforms to disclose how their recommendation algorithms work and banned AI-based price discrimination to protect consumers. These cases show industry and regulators pushing for AI that is accountable and fair.
4. How can companies design AI products ethically from the start?
Companies can adopt a “responsible by design” approach: conducting ethics reviews and bias testing during development, involving diverse stakeholders (including possibly consumer representatives) in the design process, and setting clear policies (e.g. a rule that critical decisions always get human oversight). Techniques like explainable AI, privacy-by-design (minimizing personal data use), and rigorous safety testing of AI models should be standard in the product development lifecycle.
5. What should consumers do if they suspect an AI decision is unfair or wrong?
Consumers should look for any appeal or review channels provided by the company. Many organizations are beginning to offer ways to have AI decisions reviewed by a human or to submit a complaint if an algorithm made a mistake. For example, social platforms and banks often have processes to contest content removals or credit denials. It’s also helpful for consumers to ask for an explanation of the decision – ethically run companies will be transparent and provide one. If none of these avenues exist, it may be a red flag about the company’s practices, and consumers can demand better or involve consumer protection agencies.
