- The paper presents the concept of 'decoys' as tools that misdirect accountability from the true network power structures in the AI sector.
- It employs network power analysis to demonstrate how elite actors use infrastructures, financialization, and regulatory capture to consolidate market control.
- The study advocates for reforming accountability mechanisms to target capital and data flows rather than relying on superficial technical fixes.
Reckoning with the Political Economy of AI: Beyond Decoys in Accountability
Introduction
"Reckoning with the Political Economy of AI: Avoiding Decoys in Pursuit of Accountability" (2604.16106) advances a critical theoretical and analytical intervention for AI research, policy, and the broader FAccT community. It centers the concept of the 'Project of AI' as a socio-technical endeavor embedded within, and enforcing, novel configurations of political and economic power. Moving past prevailing debates focused on technical fixes, the authors introduce the framework of "decoys"—rhetorical and institutional patterns that misdirect scrutiny from the true sources and mechanisms of networked power consolidation within the AI sector. The paper is situated at the intersection of economic sociology, STS, and communication studies, and calls for a fundamental reorientation in understanding accountability and governance in AI.
Theoretical Foundation: The AI Project as Network Power
The authors reframe prevailing analyses by foregrounding how contemporary AI is less a story of technical advances and more an ongoing project of market-making and networked power construction. Drawing from Castells’ work on network power and the actor-network theory tradition, the paper demonstrates that the foundational forces shaping AI are heterogeneous, contingent, and operate simultaneously through infrastructures (energy, data centers, chips), financialization, and strategic concert among a handful of elite actors: tech corporations, financiers, and state actors.
This network approach underscores that:
- Market formation is not natural but is actively structured by these actors, utilizing moments of uncertainty to secure and stabilize their dominance.
- Material political economy is substantiated not only in code but in data pipelines, compute infrastructure, and globalized labor, whose flows are directed—rather than regulated—by elite brokers.
- Regulatory environments are captured, with legal and supranational frameworks (e.g., regulatory proposals in the US and Europe) often co-opted as tools for market stabilization, entry barriers, and legitimation for dominant firms.
Decoys as Mechanisms of Power Preservation
A central innovation of the paper is its taxonomy of five "decoys," conceptualized as socio-rhetorical instruments that diffuse, displace, and absorb critique—paradoxically deepening networked power and market capture rather than promoting genuine accountability.
The Ontological Decoy
Debates over "what is AI" entrench the semantic ambiguity of the technology, which is exploited to expand its applicability, attract capital, and mobilize both critics and supporters into reinforcing the relevance and importance of AI as a sector. Rather than bounding the object, ontological debates serve the interests of capital by enabling flexible resource allocation, entrance of new actors, and continual narrative expansion.
The Inevitability Decoy
Framing AI's ascent as inexorable naturalizes current trajectories, delegitimizes contestation, and justifies preemptive infrastructure build-outs, risk-taking, and state interventions. It also rationalizes large capital flows and public subsidies, and positions participation in governance (e.g., stakeholder engagement, participatory design) as rituals to palliate harm, thereby diffusing resistance.
The Disruption Decoy
The recurrent assertion that AI is "disruptive" directs attention to labor automation or local sectoral shifts, obfuscating the real disruptive act: the large-scale consolidation of capital, erasure of regulatory boundaries, and restructuring of global labor into opaque, offshore, and unaccountable forms.
The Safety Decoy
"AI Safety" discourse, particularly when focused on existential risk, aligns technocratic and philosophical debate with elite interests by positioning the AI industry itself as both the source of threat and the only legitimate candidate to regulate its risks. Moral panics over AGI risk obscure immediate, distributed, and disproportionately social harms, and enable companies to reframe regulatory engagement as proactive and ethical.
The Regulatory Decoy
Where once technology incumbents lobbied against regulation, leading AI actors now call for government oversight, but with the strategic goal of shaping regulatory regimes to entrench their oligopoly. This conversion of regulatory debate into an arena for incumbents mirrors historical instances of captured regulation in other industries, turning governance into a further mechanism for market consolidation.
Implications for Accountability and Intervention
The cumulative effect of decoy logics is an accountability regime that fails to interrogate network-making, financialization, and infrastructural control at the heart of the AI sector. The paper raises the critical, and empirically robust, claim that:
- Focusing on algorithms, datasets, or even organizational leaders is structurally insufficient for accountability, as power is always redistributed, reassembled, and re-embedded at the network level.
- Accountability mechanisms must target flows: of capital, of material, of data, and of institutional influence. This entails a shift in analytic and regulatory priorities from technical objects to market and network processes.
- Standard transparency and fairness interventions risk being conscripted into the hegemonic logic of the Project of AI, serving as further decoys unless they directly address the networked configuration of power and extraction.
Anchoring Frames for AI Accountability
To combat the limitations identified above, the authors propose four anchoring frames to ground future research and policy:
- Material sites of network assembly: Examine AI infrastructure and implementation as sites of power consolidation.
- Financing as technopolitical work: Analyze the financial circuits, governance structures, and assetization that drive technical and organizational choices.
- From objects to global flows: Prioritize analyses of transnational flows of labor, capital, and computation instead of bounded technological artifacts.
- Resist social solutionism: Avoid single-discipline or technical-legal solutionism; develop transdisciplinary, network-aware practices for governance and critique.
Conclusion
The paper offers a rigorous sociological and political-economic intervention for FAccT and the broader AI governance community. By highlighting the role of decoys as constitutive—rather than incidental—mechanisms for power consolidation, it challenges actors focused on AI fairness, transparency, and accountability to critically re-examine their own practices. Genuine accountability in AI, the authors argue, can only emerge from grappling with the material, financial, and infrastructural political economy underlying the sector. Absent such an orientation, technical or regulatory strategies risk contributing to the very consolidation they hope to challenge.
This work establishes new research priorities for AI governance: empirically mapping the flows and networks that constitute the Project of AI, disambiguating the entangled interests of brokers, and devising interventions that can contest these formations, both theoretically and in practice.
