Current Affairs Topics Archive
International Relations Economics Polity & Governance Environment & Ecology Science & Technology Internal Security Geography Social Issues Art & Culture Modern History

Who are the architects of AI, and what are the ties that bind them?

February 18, 2026 5 min read Science & Technology GS3 (Science & Technology Internal Security)
On This Page ▼

What Happened

  • A close look at the global AI ecosystem reveals that a small group of researchers and entrepreneurs — often termed the "architects of AI" — simultaneously shape the technology, set safety standards, build commercial products, and advise governments
  • Key figures include Sam Altman (OpenAI), Dario Amodei (Anthropic), Demis Hassabis (Google DeepMind), and academic pioneers Geoffrey Hinton and Yoshua Bengio who now sound alarms about the technology they helped create
  • Critics warn of "industrial capture" — the risk that AI safety norms become dominated by the same companies racing to deploy the most powerful AI systems
  • The OpenAI-Anthropic rivalry has intensified over military AI contracts, with the US Pentagon awarding AI development contracts to multiple frontier labs simultaneously

Static Topic Bridges

Key AI Organizations and Their Governance Structures

The global frontier AI industry is concentrated in a handful of organisations, each with distinct ownership structures and governance philosophies that have significant implications for safety and public accountability.

  • OpenAI: Originally a non-profit, restructured as a "capped-profit" company; CEO Sam Altman; raised over $150 billion in total funding; removed explicit bans on military use from its policies in 2024
  • Anthropic: Founded in 2021 by former OpenAI employees including Dario and Daniela Amodei; structured as a "Public Benefit Corporation"; focuses on Constitutional AI and safety research
  • Google DeepMind: Formed by merger of Google Brain and DeepMind (2023); CEO Demis Hassabis; operates within Alphabet's corporate structure; developed the Frontier Safety Framework
  • Meta AI: Led by Yann LeCun; open-source approach (LLaMA models publicly released) — contrasts with closed approaches of OpenAI and Anthropic
  • The "big three" (OpenAI, Anthropic, Google DeepMind) received Pentagon contracts worth up to USD 200 million each to prototype frontier AI for national security purposes

Connection to this news: The concentration of frontier AI development in a few interlinked organizations — whose founders and key personnel move between them — raises governance questions analogous to regulatory capture in finance.

AI Safety: Academic Concerns vs. Commercial Reality

The AI safety movement began as an academic concern about existential risks from advanced AI (also called "AGI risk") but has now been partially co-opted by commercial labs as a branding and regulatory strategy. The tension between genuine safety research and commercial deployment is a defining feature of the current AI landscape.

  • Geoffrey Hinton (Nobel Prize in Physics 2024 for contributions to AI): Resigned from Google in 2023 to speak freely about AI risks; warns about loss of control over AI systems and technological unemployment
  • Yoshua Bengio (Turing Award 2018, shared with Hinton and LeCun): Has signed open letters calling for a pause on superintelligence development; testified before US Senate on AI risks
  • Jan Leike: Former OpenAI alignment lead who resigned in 2024, writing that "safety culture and processes have taken a backseat to shiny products" at OpenAI before joining Anthropic
  • Voluntary safety frameworks (OpenAI's Preparedness Framework, Anthropic's Responsible Scaling Policy, Google DeepMind's Frontier Safety Framework) are industry-self-regulated — no independent verification mechanism
  • The 2023 Bletchley Park AI Safety Summit and 2024 Seoul Summit established inter-governmental dialogue on AI safety but produced no binding treaty

Connection to this news: The question of who governs AI development — and whether the same people building it can objectively assess its risks — is central to UPSC themes of technology governance, regulatory frameworks, and accountability.

India's AI Governance Framework

India has adopted a principles-based, innovation-friendly approach to AI governance rather than prescriptive regulation, positioning itself as a model for the Global South.

  • India AI Governance Guidelines released in November 2025 by the Ministry of Electronics and Information Technology (MeitY) — principles-based, not binding law
  • India's national AI strategy "AI for All" (2018) framed by NITI Aayog; updated as IndiaAI Mission (2024) with ₹10,372 crore outlay
  • IndiaAI Mission pillars: compute infrastructure, data platform, application development, future skills, startup ecosystem, safe & trusted AI, research
  • India's approach: "light-touch, agile, flexible" regulation for general use; targeted intervention for specific harms (deepfakes, election interference)
  • India is a member of the Global Partnership on AI (GPAI) and has co-led the AI track at G20 (New Delhi Presidency, 2023)
  • Key regulatory gap: No binding AI Act equivalent to the EU AI Act (world's first binding AI law, in force 2024)

Connection to this news: As the architects of AI concentrate in a few Western firms with government contracts, India's choice of governance model — whether to regulate, partner, or build domestic alternatives — is a critical strategic question.

Large Language Models (LLMs) and Generative AI — Technical Concepts for Prelims

Large Language Models are a class of AI systems trained on vast amounts of text data using a technique called "transformer architecture" (introduced by Google researchers in the seminal 2017 paper "Attention Is All You Need"). They power chatbots, code generators, and content tools.

  • LLMs work via next-token prediction — they learn statistical patterns in language at massive scale
  • "Alignment problem": ensuring AI systems pursue goals intended by their designers is the core technical challenge in AI safety
  • Constitutional AI (Anthropic's method): AI models are trained using a set of principles to self-critique and revise outputs
  • RLHF (Reinforcement Learning from Human Feedback): Used by OpenAI, Anthropic, and others to make AI outputs more helpful and less harmful
  • The EU AI Act (2024) classifies AI systems by risk level — "unacceptable risk" (banned), "high risk" (regulated), "limited risk" (transparency obligations), "minimal risk" (no obligation)
  • India does not yet have a comparable risk-classification framework for AI

Connection to this news: Understanding how LLMs work and who governs them is essential context for UPSC questions on AI policy, digital sovereignty, and emerging technology regulation.

Key Facts & Data

  • OpenAI founding: 2015 (non-profit), restructured as capped-profit entity; Sam Altman is CEO
  • Anthropic founding: 2021 by Dario and Daniela Amodei (former OpenAI); structured as Public Benefit Corporation
  • Google DeepMind: Formed 2023 (merger of Google Brain + DeepMind); Demis Hassabis is CEO
  • Geoffrey Hinton: Nobel Prize in Physics 2024; resigned from Google 2023 to speak on AI risks
  • Yoshua Bengio: Turing Award 2018 (shared with Hinton and LeCun); calls for superintelligence development pause
  • US Pentagon AI contracts: Up to USD 200 million each to OpenAI, Anthropic, Google DeepMind, xAI
  • IndiaAI Mission budget: ₹10,372 crore (approved 2024)
  • EU AI Act: In force 2024 — world's first binding AI law; uses risk-classification framework
  • Bletchley Park AI Safety Summit: November 2023; Seoul AI Summit: May 2024
  • Key doctrine: "Industrial capture" — concern that AI safety is dominated by the same firms racing to deploy AI