Bias, surveillance and job loss: AI’s hidden costs in circular economy

Circular economy initiatives often rely on data-intensive, AI-driven technologies to optimize waste sorting, material reuse, product lifecycle management, and circular service platforms. Yet, AI’s opaque “black box” decision-making introduces severe transparency issues. The study emphasizes that stakeholders in circular supply chains, such as recyclers, consumers, and regulators, may struggle to understand or challenge AI-driven outcomes without clear algorithmic explainability. This absence of clarity poses barriers to accountability and trust, particularly in high-impact areas like material sorting or dynamic pricing of reused goods.

A new scholarly review warns businesses, governments, and technology developers that without ethical safeguards, artificial intelligence (AI) may undermine rather than accelerate the circular economy (CE). The review, titled "Ethical Aspects of AI Use in the Circular Economy", was published in the journal AI & Society.

The study examines over 950 sources, peer-reviewed papers, policy reports, and industry case studies, detailing how ethical risks linked to AI could block sustainable development unless addressed through targeted governance and inclusive design.

What are the key ethical concerns linking AI to circular economy systems?

The study identifies six primary ethical risk areas, algorithmic transparency, data privacy, bias, labor impacts, social inclusion, and governance, each intensified by the unique structure of circular economic models.

Circular economy initiatives often rely on data-intensive, AI-driven technologies to optimize waste sorting, material reuse, product lifecycle management, and circular service platforms. Yet, AI’s opaque “black box” decision-making introduces severe transparency issues. The study emphasizes that stakeholders in circular supply chains, such as recyclers, consumers, and regulators, may struggle to understand or challenge AI-driven outcomes without clear algorithmic explainability. This absence of clarity poses barriers to accountability and trust, particularly in high-impact areas like material sorting or dynamic pricing of reused goods.

Moreover, the integration of AI introduces expansive surveillance risks. Circular systems increasingly depend on IoT sensors, smart bins, and usage-monitoring devices, raising questions over how personal data are collected, stored, and used. The study notes examples where recycling habits or energy consumption patterns inadvertently revealed private lifestyle data, without user consent or awareness. Without privacy-by-design protections, these technologies may infringe on individual rights and provoke public resistance.

Algorithmic bias is another major ethical flashpoint. AI trained on unrepresentative or skewed datasets can disproportionately disadvantage specific communities in circular programs. Examples include pricing disparities in fashion resale platforms or inequitable access to smart recycling services in low-income neighborhoods. Such bias not only undermines fairness but may exacerbate existing social inequalities - an outcome fundamentally at odds with the inclusive vision of circularity.

The research also explores how AI reshapes the labor landscape. While AI can increase efficiency in recycling and remanufacturing, it risks automating away low-skill jobs without proper reskilling pathways. Informal waste workers in developing nations are especially vulnerable, potentially displaced by AI-powered systems developed and deployed in high-income countries. These labor disruptions could provoke backlash unless transitions are managed with ethical foresight.

Social inclusion is similarly threatened when AI-based CE solutions assume universal digital access. Many elderly or economically disadvantaged populations may lack the infrastructure or literacy to participate in smart CE programs, such as app-based recycling rewards. Without inclusive design alternatives, these populations risk exclusion from key sustainability benefits.

How do these ethical challenges affect real-world adoption of circular economy practices?

The study highlights a critical point: ethical risks are not just abstract concerns - they directly influence whether AI-based circular solutions are accepted, trusted, and scaled in the real world.

Case studies featured in the review vividly illustrate this dynamic. In one European city, smart recycling bins equipped with AI initially excluded elderly and low-income residents due to digital barriers. Public outcry led city officials to introduce physical access cards, data transparency statements, and digital literacy campaigns - interventions that later increased trust and participation.

In the corporate sphere, a fashion resale platform discovered that its pricing algorithm undervalued items from certain neighborhoods. By removing location-based pricing factors and increasing transparency through user-facing pricing breakdowns, the company not only corrected bias but also regained user trust.

Labor concerns are also addressed through concrete examples. Apple’s AI-powered recycling robot “Daisy” significantly enhances material recovery but risks displacing human recyclers in the Global South. The review notes that Apple's offer to license the technology and potential integration of displaced workers into formal recycling facilities represent attempts, albeit partial, to balance productivity with social equity.

These cases demonstrate that ethical missteps can rapidly erode stakeholder support, hinder user engagement, and attract regulatory scrutiny. In contrast, ethical responsiveness, through inclusive design, human oversight, bias audits, and participatory governance, has been shown to improve adoption and societal acceptance.

What governance and policy tools can ensure AI supports a fair circular transition?

In a nutshell, the study asserts that responsible governance is essential to ensure AI functions as an enabler, not an obstacle, for circular economy goals.

Global frameworks such as the OECD AI Principles and the EU AI Act are identified as crucial starting points. These documents promote values of transparency, accountability, fairness, and human oversight. Yet, the study warns that generic ethical frameworks alone are insufficient for the circular context, where AI decisions may affect environmental outcomes, material flows, and public infrastructure.

To bridge this gap, the review recommends domain-specific ethical guidelines tailored to CE challenges, especially concerning traceability in industrial symbiosis, lifecycle accountability of AI-driven products, and equitable data governance in collaborative platforms.

Among the actionable strategies proposed are:

• Algorithmic transparency mandates, requiring explainable models or at minimum traceable documentation for auditability.
• Data privacy protections, through techniques like federated learning, anonymization, and user consent mechanisms.
• Bias mitigation measures, including representative datasets, fairness constraints, and independent algorithmic audits.
• Labor transition plans, with proactive investment in workforce reskilling and inclusion of informal sector workers in new value chains.
• Inclusive system design, offering non-digital alternatives and accessibility features for marginalized communities.
• Human-in-the-loop governance, ensuring that critical AI decisions remain subject to human review and override in CE infrastructures.

The study also warns of a growing risk of "AI greenwashing", where companies overstate sustainability benefits without rigorous evidence. Responsible governance frameworks should include mechanisms to validate AI's environmental claims, avoid superficial sustainability branding, and foster transparent impact assessments.