Chris Olah, co-founder of AI safety company Anthropic, has argued that artificial intelligence development and governance cannot remain concentrated within the hands of wealthy technology corporations if the world expects equitable distribution of AI’s benefits across nations and societies. Speaking on the structural imbalances in global AI advancement, Olah contended that external oversight mechanisms and guidance frameworks are essential to prevent a handful of private corporations from controlling the trajectory of transformative technology.
The concern reflects a growing anxiety within AI research circles about the centralisation of computational power, talent, and decision-making authority among a small cluster of well-capitalized tech firms—primarily based in the United States and China. Anthropic itself, founded in 2021 by former OpenAI researchers including Olah, has positioned itself as an alternative to larger players, emphasizing AI safety and interpretability. Yet even Anthropic operates within the broader ecosystem where computational resources, investment capital, and engineering talent concentrate among Silicon Valley and Beijing-based entities. This concentration creates what experts term a “power asymmetry”—where decisions made by corporate research teams in San Francisco or Shenzhen ripple across global economies with limited democratic input or regulatory oversight.
The challenge gains particular urgency for India and South Asia, where AI adoption is accelerating but indigenous capacity for AI research and development remains limited compared to the West and China. Indian technology companies and startups increasingly depend on foundational models developed by US corporations like OpenAI, Google DeepMind, and Meta. This dependency creates a strategic vulnerability: when AI governance decisions are made exclusively by American or Chinese corporations, emerging economies have minimal leverage in shaping how these systems function, what biases they embed, or how their benefits are distributed. India’s own AI initiatives—including government-backed research programmes and corporate ventures—operate in this constrained landscape.
Olah’s proposal for external governance suggests establishing independent bodies with authority to guide AI development standards, safety protocols, and access frameworks. Such mechanisms might include international regulatory bodies, academic consortiums, or multi-stakeholder governance models that include representation from developing nations. The rationale is straightforward: technology with global impact should have global input into its governance. Currently, a startup founder in Bangalore building AI applications must work within guardrails set by distant Silicon Valley firms; a researcher in Mumbai cannot meaningfully influence how training datasets are constructed or what ethical guidelines are embedded in foundational models.
India’s technology sector—encompassing software services giants like TCS, Infosys, and Wipro, alongside emerging AI startups—faces a dual imperative. These firms must remain competitive by adopting cutting-edge AI capabilities developed elsewhere, yet doing so entrenches dependence on external technology providers. The Indian government, meanwhile, has outlined ambitions to position India as an AI hub through initiatives like the National AI Strategy, yet this strategy unfolds within parameters largely set by external corporate actors. Olah’s argument implicitly suggests that such dependencies are unsustainable long-term and that genuine equity requires moving beyond corporate benevolence toward structured external governance.
The broader implication extends beyond technology policy into questions of economic sovereignty and geopolitical alignment. Which nations set AI standards becomes consequential: standards favouring certain languages, cultural contexts, or regulatory philosophies will advantage some economies over others. If AI governance remains corporate-led and Western-centric, South Asian nations risk embedding technological dependencies that constrain their long-term development optionality. Conversely, overly fragmented governance—where nations pursue isolationist AI policies—could splinter the technology ecosystem and reduce innovation velocity globally. The optimal pathway likely involves hybrid models where external bodies establish baseline standards while allowing regional and national flexibility in implementation.
Looking forward, Olah’s intervention signals that AI governance will become increasingly contested terrain. The coming years will determine whether corporations maintain primary authority over AI development or whether independent oversight bodies gain meaningful power. For India and South Asia, the stakes are tangible: access to AI capabilities, voice in setting technical standards, and the ability to shape how these systems function within their societies. Whether external governance mechanisms materialise, and whether they include adequate representation from emerging economies, will shape whether AI’s benefits genuinely distribute globally or concentrate further among wealthy Western and Chinese entities.
