In our previous AI and CSR Series entries on labour practices, human rights, and organizational governance, we explored how AI impacts each pillar of corporate social responsibility. Now, in Entry #6, we turn to ISO 26000’s concept of Fair Operating Practices – essentially, the ethical conduct of an organization in its dealings with other businesses, government, and society. ISO 26000 emphasizes fair competition and preventing corruption as foundations of ethical business conduct. In practice, this means companies should compete honestly, comply with laws, reject bribery, and be transparent about their business practices.
But how do these principles hold up in the era of AI? Artificial intelligence is rapidly being embedded into business operations – from algorithmic pricing in e-commerce to AI-driven decision-making in finance and procurement. This raises new questions about AI compliance (are algorithms obeying the law and regulations?), AI ethics (do AI systems align with our ethical principles and CSR commitments?), and potential pitfalls like algorithm-driven price fixing or automated fraud.
Transparency and accountability are core to both ISO 26000 and responsible AI. ISO 26000’s guiding principles include transparency and respect for the rule of law – organizations must openly communicate about their systems and obey all applicable laws. With AI, ensuring transparency might involve disclosing how an algorithm makes decisions or auditing AI outcomes for bias. Respecting the law means that AI systems should be designed not to inadvertently break competition law, privacy law, or anti-bribery statutes. As ISO 26000 puts it, ethical business conduct “goes beyond legal compliance” but never falls below it.
In this article, we examine how AI is testing and transforming fair operating practices in three key areas – fair competition, anti-corruption, and business ethics – and what organizations are doing (especially in Asia and Singapore) to harness AI for good governance rather than abuse it. We’ll highlight real-world cases from 2023 onwards, from Singapore’s use of AI to detect bid-rigging to global antitrust actions on algorithmic pricing. Throughout, we consider how businesses can pursue responsible innovation – leveraging AI’s benefits while upholding ethical AI principles and ISO 26000’s guidance on fairness and integrity.
(AI and CSR Series – see earlier entries for AI’s impact on labor, human rights, environmental responsibility, and governance.)
Aligning AI with business ethics is a critical part of fair operating practices. ISO 26000 calls for ethical behavior in business dealings – meaning honesty, integrity, and consideration of stakeholders. Ethical AI use is now an extension of this principle. Companies deploying AI should ensure these systems reflect their code of ethics and do not engage in behavior that a human would deem unethical.
One major concern is transparency in AI decision-making. If an AI system recommends denying a loan or selecting a vendor, can the company explain why? Black-box algorithms pose an ethical risk by obscuring the rationale behind decisions, potentially hiding bias or unfair treatment. This conflicts with ISO 26000’s emphasis on transparency and accountability. In response, organizations and governments are developing AI governance frameworks to inject transparency and fairness into AI. For example, Singapore’s Model AI Governance Framework (updated in 2024) centers on principles that AI decisions should be explainable, transparent, and fair. Businesses are encouraged to make their AI models interpretable and to communicate clearly about how AI is used in operations. Such measures help maintain stakeholder trust – a key CSR goal – by showing that AI is not a mysterious unruly force but subject to oversight and ethical guidelines.
Another aspect of AI ethics in business is avoiding biases or discrimination. An AI system might unintentionally replicate unfair biases (e.g. in hiring or supplier selection) if trained on biased data. This violates the spirit of fair operating practices, which include treating partners and stakeholders fairly and without discrimination. Companies in Asia are actively addressing this: Singapore, for instance, has made AI ethics “a commitment to fairness and transparency,” integrating it into regulatory expectations. Firms are advised to vet their AI tools for bias and ensure decisions are fair and equitable. This is not just a moral stance but also reduces legal risks – an AI that discriminates could lead to lawsuits or regulatory penalties.
Responsible innovation comes into play here. Businesses should innovate with AI in ways that do not sacrifice ethical standards. This concept is gaining traction in Asia-Pacific. Vietnam established an AI Ethics Committee to promote “safe AI – shaping responsible innovation”, and Hong Kong issued guidelines for responsible innovation to ensure generative AI is deployed ethically in finance. The idea is to proactively govern AI rather than react after harm occurs. As a practical example, Singapore uses regulatory sandboxes for AI – controlled environments where new AI solutions can be tested with oversight. These sandboxes allow firms to experiment with AI-driven services (for instance, in fintech or healthcare) while regulators monitor outcomes. This approach encourages creativity and competitiveness through AI, but with safeguards so that if an AI starts behaving in a problematic way, it can be corrected before full rollout. It’s a balance of fostering innovation and enforcing ethics – the essence of responsible AI deployment.
In summary, aligning AI with business ethics means embedding values like transparency, fairness, and accountability into AI systems. Companies are writing AI ethics charters and training employees on ethical AI use. Some have internal AI ethics boards to review high-risk AI projects. These steps echo ISO 26000’s guidance that an organization’s culture of integrity must extend to all tools it uses – including algorithms. By treating AI outcomes with the same scrutiny as human decisions, businesses uphold their CSR commitments even as they adopt advanced technology.
Ensuring fair competition is a cornerstone of fair operating practices. This means businesses compete on merit – price, quality, innovation – rather than through manipulation or anti-competitive agreements. However, the rise of AI-driven pricing and market analytics has introduced new challenges. Specifically, regulators worldwide are wary of algorithmic collusion: the risk that autonomous pricing algorithms could coordinate prices or market behavior in a way that reduces competition, even without explicit human agreements.
Consider modern pricing algorithms used by airlines, ride-hailing apps, or e-commerce platforms. These AI systems adjust prices dynamically based on supply, demand, and competitor prices. If multiple competitors deploy similar algorithms, there is a concern that prices could converge at higher levels, effectively fixing prices above competitive rates. In antitrust terms, the algorithms might achieve a tacit collusion outcome – all without direct communication between company executives. This scenario isn’t just theoretical. In late 2024, the U.S. Department of Justice filed a landmark case against RealPage, a software firm whose AI-driven platform was used by major landlords to set rental prices. The DOJ alleges that RealPage’s algorithm collected confidential pricing data from competing property owners and then recommended rent prices that eliminated competitive pressure, leading to uniformly higher rents. In essence, the AI became a hub for a price-fixing scheme, albeit one that users may have viewed as mere “analytics.” This case – still ongoing as of 2025 – demonstrates regulators’ resolve to treat certain algorithm-driven coordination as cartel behavior, no different from human collusion.
In Asia, algorithmic collusion has also been identified as an emerging risk, even if concrete cases have yet to surface publicly. Singapore’s Competition and Consumer Commission (CCCS) noted as early as 2020 that widespread use of AI in pricing “could increase the risk of collusion between competitors on digital platforms.”. Fast forward to 2024, and CCCS is proactively developing tools to address this issue. The agency’s chief executive highlighted a new AI governance initiative: an extension to the AI Verify toolkit that will let companies self-assess their AI pricing systems for potential anti-competitive behavior. This tool (a first of its kind) essentially allows businesses to test if their algorithms might be unintentionally recommending collusive prices or biased outcomes. By using it, a company could catch a problem like an algorithm that consistently matches a competitor’s price increases – and then tweak the AI to behave more competitively. Such preventive measures reflect a novel approach to competition law in the AI era: rather than waiting to punish collusion after the fact, authorities want to build compliance into algorithm design from the start.
Beyond pricing, AI can affect competition in other ways. Recommendation engines could unfairly prefer one’s own products over a competitor’s (think of a marketplace platform secretly boosting its in-house brands in search results). Or large AI-driven platforms might engage in exclusionary practices, like using algorithms to block interoperability with rivals (a concern raised in China, where regulators banned using algorithms to create “walled gardens” that lock out competitors). Additionally, the concentration of AI resources – such as only a few big tech firms controlling advanced AI models or critical data – is being watched closely. Singapore’s competition authority pointed out the risk of emerging monopolies in AI, since developing cutting-edge models requires huge datasets, specialized chips, and cloud computing power that only a handful of giants possess. If left unchecked, this could lead to dominance over the AI ecosystem, stifling competition and innovation.
To uphold fair competition, regulators in Asia-Pacific are updating policies and guidance. China amended its rules in 2022 to explicitly prohibit use of algorithms for price manipulation and collusion, defining coordinated use of identical pricing algorithms among competitors as illegal. The new Chinese provisions also mandate transparency about algorithms’ “fundamentals and mechanisms” to regulators– aiming to prevent secretive algorithmic tactics. This mirrors ISO 26000’s call for transparency and also ties into corporate social responsibility: interestingly, Chinese rhetoric framed algorithmic price-fixing as an affront to consumers’ rights and corporate social responsibility. Meanwhile, competition agencies in jurisdictions like Japan, South Korea, and the EU are studying algorithmic collusion cases and warning companies that blaming “the machine” is no defense if their AI breaks antitrust laws.
Case in point: Even if businesses do not intend to collude, they could be held liable if their algorithms do so. Therefore, AI compliance programs now must include antitrust compliance for algorithms. Companies are beginning to train their data science and pricing teams on competition law basics – ensuring that they program or tune AI in ways that promote independent decision-making. Technical measures can help, such as adding randomness in pricing algorithms or avoiding algorithms that rely on competitors’ non-public data. The bottom line is that fair competition must be encoded into the AI systems that drive market behavior. By doing so, companies not only avoid hefty fines and reputational damage but also contribute to a healthy market ecosystem, aligning with the broader CSR goal of sustainable, fair markets.
Anti-corruption – particularly anti-bribery – is another pillar of fair operating practices under ISO 26000. Organizations are expected to work against corruption in all forms, including extortion and bribery. AI technologies are becoming both tools and targets in the fight against corruption. On one hand, AI can significantly enhance compliance programs by detecting fraud and flagging unethical conduct. On the other hand, if misused, AI could also enable new forms of wrongdoing or obscure old ones.
Let’s look at the positive side first. Companies and governments are increasingly deploying AI to detect irregularities that might indicate fraud or corruption. Machine learning systems excel at sifting through huge datasets – far more than any human auditor could – and spotting patterns or anomalies. In procurement and finance, this has been a game changer. A noteworthy example comes from Singapore’s public sector: the national R&D agency ASTAR developed an AI tool specifically to predict and prevent procurement fraud. This system analyzes a mix of data – HR records, finance transactions, procurement requests, tender approvals, and even relational data – to identify red flags. For instance, it can correlate employee data with vendor records to highlight suspicious links, such as if a government official has family members who work for a supplier that keeps winning bids. It also notices unusual purchasing patterns (e.g. one employee consistently approving purchases of a certain product, which might indicate favoritism or kickbacks). By running these analyses regularly, the tool helps compliance officers catch early signs of corruption or conflict of interest before they escalate into scandals. In fact, ASTAR has been using this system internally to safeguard its own purchasing, and it was trialed across multiple Singapore agencies as a proactive anti-graft measure. Such AI-driven oversight builds on the transparency and accountability ethos of ISO 26000 – using data to ensure openness and fair play in organizational processes.
Similarly, AI is assisting in financial compliance. Banks in Asia employ AI to monitor transactions for signs of money laundering or bribery payments (for example, unusual payment patterns that match known typologies of illicit payments). These AI compliance tools can cross-reference numerous data points – amounts, timing, counterparties, even text in payment memos – far faster than traditional rule-based systems. The goal is to catch bribery or embezzlement attempts that human reviewers might miss. In the construction industry, AI has been used to analyze project data to find inconsistencies that could indicate fraud or cost inflation. All these applications underscore how, used responsibly, AI can strengthen internal controls and uphold an organization’s commitment to anti-corruption.
Now the flip side: AI can also pose new corruption risks if ethics and oversight are lacking. One concern is that AI systems might inadvertently mask corrupt behavior. For example, if a procurement algorithm is trained on historical data that unfortunately contains instances of favoritism, the AI might “learn” to continue favoring certain suppliers (e.g. always awarding contracts to those with past wins, which might include those who secured deals via bribes). Without careful checks, the AI could perpetuate a cycle of corruption under the guise of efficiency. This is why respect for the rule of law and anti-bribery principles must be encoded into AI systems – e.g. by excluding tainted historical data or including rules that flag rather than accept patterns consistent with bid-rigging or collusion.
Another emerging threat is the use of AI-generated deepfakes and automation in fraud. While not classic bribery, these are corrupt acts in a broader sense and can impact businesses. For instance, there have been cases in Hong Kong and elsewhere where criminals used AI-generated voices or videos to impersonate CEOs and authorize fraudulent payments – essentially high-tech social engineering. A Hong Kong company in 2023 was swindled out of US$25 million through a deepfake video call imitating its CFO. Although this is external crime, it exposes companies to huge losses and legal trouble (imagine if that money was misappropriated from client funds). It pressures companies to beef up their verification processes (e.g. multi-factor authentication for approvals) as part of compliance. Additionally, AI can assist corrupt insiders: an employee might use generative AI to automatically produce fake invoices or forge documents at scale, overwhelming traditional controls. Compliance teams thus have to stay one step ahead, possibly using AI themselves to verify authenticity of documents and communications.
Governments in Asia are recognizing these challenges. Anti-corruption agencies are exploring AI for investigations – scanning emails for bribery keywords or analyzing public procurement data for bid-rigging cartels. At the policy level, there’s movement towards clearer guidance. International standards like ISO 37001 (on anti-bribery management systems) complement ISO 26000 by giving organizations a framework to prevent, detect, and respond to bribery. While not AI-specific, such standards increasingly acknowledge technology’s role. For example, a robust compliance system under ISO 37001 would include continuous monitoring – something AI can excel at.
Ultimately, maintaining fair operating practices in the age of AI means updating our anti-corruption toolkit. Companies should integrate AI into their compliance programs – but with adequate human oversight (the “human-in-the-loop” approach) to verify AI findings and avoid false positives or negatives. Training programs for employees should now cover digital ethics: e.g. warning procurement officers that using an AI assistant doesn’t absolve them from due diligence (you can’t blame a biased vendor choice on “the AI said so”). Likewise, strong governance is needed so that any AI system that handles financial or procurement tasks is reviewed by compliance or audit teams. By harnessing AI’s data-crunching power while keeping ethical guardrails, organizations can enhance transparency and root out corruption more effectively than before – living up to the anti-corruption principle of ISO 26000 in a high-tech environment.
A recurring theme in these discussions is balance – specifically, balancing innovative use of AI with the responsibility to uphold ethical standards and comply with the law. This is the essence of responsible innovation. Companies do not want to stifle the creative potential of AI (which can drive efficiency, customer value, and growth), but they also cannot afford to ignore the compliance and ethical implications. How can organizations strike this balance?
One strategy is building a strong AI governance framework internally. Many forward-looking companies in Asia and globally have instituted AI governance committees that include members from compliance, legal, IT, and business units. These committees evaluate proposed AI deployments (say, a new AI tool for recommending retail investment products or an AI chatbot interfacing with customers) for risks and alignment with company values. Questions they consider: Does the AI comply with data protection laws? Could it inadvertently discriminate or give unfair outcomes? Is there a process to monitor its decisions and override or fix them if something goes wrong? By vetting AI projects in this cross-functional way, businesses infuse compliance and ethics into innovation from day one.
Another key is ongoing monitoring and audits of AI systems. Responsible innovation acknowledges that you can’t just set an AI loose and hope for the best. Instead, companies are establishing continuous oversight – for example, periodic audits of an algorithm’s outputs to ensure they remain fair and legal. If an audit finds an anomaly (maybe a pricing AI started suggesting unusually high prices for certain segments), the company can pause and adjust the system. This echoes the “plan-do-check-act” approach common in quality management and compliance: innovate and deploy, but also regularly check and improve.
Regulators are also signaling that compliance enforcement in digital services will intensify. In other words, just because a service is delivered via an algorithm or digital platform doesn’t exempt it from the rules. For instance, Singapore’s MAS (Monetary Authority) has made clear that financial institutions using AI for customer service or trading must still adhere to conduct standards – misleading advice from a robo-advisor is as unacceptable as from a human advisor. We see sector-specific guidelines emerging, like in healthcare where AI diagnostic tools should meet medical ethics and safety regulations, or in transportation where autonomous vehicles must follow traffic laws and safety norms. These all feed into a broader point: compliance in the AI era may require new techniques (like code audits, algorithmic impact assessments), but the fundamental expectation remains that businesses control their processes and outputs, AI-driven or not.
Importantly, responsible AI innovation is increasingly viewed as a competitive advantage, not a burden. Companies that manage to innovate within ethical boundaries tend to earn greater public trust. In fields like fintech or e-commerce, trust can translate to user adoption and brand loyalty. Moreover, being proactive can pre-empt heavy-handed regulation. ASEAN, for example, released an AI governance guide for businesses to follow voluntarily, promoting a culture of responsibility that could obviate the need for strict laws down the line. Singapore’s approach to AI regulation has been to avoid broad-brush laws and instead encourage industry to adopt best practices for ethical AI and data governance, thereby “ensuring responsible innovation.
We should also note the role of stakeholder engagement in responsible innovation. ISO 26000 highlights respect for stakeholder interests and community involvement. In the AI context, this means companies should listen to concerns from customers, employees, and society when rolling out AI innovations. For instance, if users are uncomfortable with how an AI uses their data, a responsible innovator will adjust and perhaps offer opt-outs, rather than pushing forward recklessly. This collaborative approach can uncover ethical issues early and generate solutions that pure internal brainstorming might miss.
In conclusion of this section and the article: AI and fair operating practices are not at odds – they can be mutually reinforcing if managed well. AI can advance compliance (through better monitoring and analytics) and make competition fairer (through tools that check anti-competitive tendencies), and it can certainly help eradicate corruption (through diligent oversight). But achieving these benefits requires a conscious effort to integrate CSR principles into AI strategy. Fair operating practices in the age of AI demand that organizations be as innovative in their governance as they are in their technology. Asia’s experience, from Singapore’s AI ethics frameworks to China’s algorithmic laws, shows a path forward where AI is harnessed in service of fair, transparent, and ethical business. For companies everywhere, the mandate is clear: embrace AI’s opportunities, but do so with eyes open to the risks, a compass firmly set on ethical principles, and a commitment to fair and responsible innovation.
(This article is part of the “AI and CSR Series,” examining how AI intersects with corporate social responsibility topics. Previous entries have covered AI’s relationship with labor rights, human rights, organizational governance, and environmental responsibility. Future entries will continue to explore the remaining facets of CSR in the AI context.)
Q1. What are “fair operating practices” in ISO 26000?
A1. Fair operating practices refer to ethical conduct in an organization’s dealings with other entities – this includes obeying laws, avoiding corruption and anti-competitive behavior, being transparent, and promoting social responsibility in its sphere of influence. Essentially, it’s about doing business fairly and with integrity.
Q2. What does AI compliance mean for businesses?
A2. AI compliance means ensuring that AI systems and their outcomes adhere to all relevant laws, regulations, and ethical standards. This involves monitoring AI decisions for issues like bias or collusion, complying with data privacy rules, and following industry guidelines so that the use of AI does not lead to legal violations or unethical outcomes.
Q3. What is algorithmic collusion and why is it a concern?
A3. Algorithmic collusion is when pricing algorithms or AI systems used by different companies effectively coordinate to keep prices high or otherwise reduce competition (intentionally or unintentionally). It’s a concern because it can lead to price-fixing or cartel-like outcomes without an explicit agreement – harming consumers and violating competition (antitrust) laws.
Q4. How can AI help in anti-corruption efforts?
A4. AI can be a powerful tool against corruption by analyzing large datasets to detect patterns and anomalies that humans might miss. For example, AI can flag unusual transactions or relationships in procurement that suggest bribery or fraud. It can also monitor communications for red flags. In short, AI can enhance compliance teams’ ability to identify and prevent corrupt practices early.
Q5. How is Singapore addressing AI ethics and responsible AI use?
A5. Singapore has taken a proactive approach: it introduced the Model AI Governance Framework to guide businesses in ethical AI deployment, emphasizing transparency, fairness, and accountability. The government also launched AI Verify, a toolkit for testing AI systems for issues like bias or collusion. Regulators encourage responsible innovation by using sandboxes for AI experiments and have signaled that existing laws (e.g. on consumer protection and fair competition) will be enforced equally on AI-driven services.
