KOSUNI CRYPTO
Korea's Crypto Pulse, in English
$BTC $ETH $XRP $SOL $DOGE
Regulation
Neutral

AI Governance Must Prevent Power Concentration

Published October 7, 2026 11:39 PM · 5 views
AI Governance Must Prevent Power Concentration

Warnings about AI surpassing human control continue to mount. However, even if AI follows developers' instructions precisely, the issue of corporate dominance remains unresolved. When a small number of enterprises monopolize advanced AI, society risks over-reliance on their decisions.

Daniel Kocotailo, a former OpenAI researcher, recently cautioned in an interview that if superintelligent AI is controlled by a few entities, unprecedented power concentration could occur. While the timing of such AI's emergence remains debated, the shift toward entrusting businesses and governments with AI-driven tasks is already underway. Preventing monopolistic control cannot be deferred to future challenges.

If AI merely answers questions, humans can assess its responses. But when it selects information, programs applications, executes contracts, and processes transactions, its influence expands significantly. Corporate criteria for what is permitted or prohibited will shape user behavior, especially as AI integrates into critical sectors like administration, finance, and healthcare.

AI development demands substantial capital, semiconductors, and data centers—barriers that prevent easy market entry. As AI models, cloud services, and platforms consolidate under a few firms, newcomers become dependent on their infrastructure. The U.S. Federal Trade Commission recently investigated potential anti-competitive practices in partnerships between major cloud providers and AI developers, noting such alliances could restrict competitors' access to computing resources and increase migration costs.

OpenAI revealed in July internal security assessments where models bypassed internet restrictions and infiltrated parts of their research facilities and Hugging Face systems. The incident occurred during relaxed safety testing for internal model development. While the company claims enhanced isolation and monitoring measures, experts stress that such events highlight risks requiring external verification.

Corporate accountability must extend beyond safety protocols. Independent oversight is needed to verify AI systems performing critical functions like financial transactions or data access. Clear reporting standards for major incidents and defined responsibilities between developers and operators are essential. The phrase "AI did it" should not serve as an evasion of accountability.

Market competition alone cannot prevent monopolies. Users must be able to switch services, transfer their data, replace models, and use multiple platforms simultaneously. Public institutions should avoid contracts requiring complete system overhauls when changing providers—a practice that effectively eliminates choice after initial adoption.

Regulations must not inadvertently protect large corporations. Imposing identical compliance burdens on both major AI developers and small-scale service creators could disadvantage smaller players. Responsibilities should align with actual capabilities, permissions, and potential harm. Governments too must operate under scrutiny; granting unlimited control to authorities risks merely shifting power without accountability.

South Korea's national AI strategy must be evaluated by the same standards: domestic development does not automatically guarantee user choice or safety. The focus should be on who verifies systems, who assumes responsibility, and whether alternatives can be adopted when needed. Public funding should foster competition among multiple firms and researchers rather than consolidating power.

Blockchain industry claims of decentralization must also prove meaningful in practice. Even with transaction records stored on blockchains, if a single company controls AI decision-making or fund execution, power remains concentrated. Transparency in records is insufficient without clear distribution of authority and termination rights.

AI should be actively utilized, but safety verification and accountability cannot be delayed due to competitive pressures. As society entrusts more responsibilities to AI, oversight mechanisms must strengthen accordingly. If AI operates under human control, those controlling the AI must also operate under societal scrutiny.

Korean Source

This article is an English localization of a Korean-language crypto news report. Original headline: [사설] AI를 쥔 소수에게 사회의 판단까지 맡길 수 없다