When a technology company's product design becomes de facto public policy, who should be accountable? Discuss with reference to AI in classrooms.

When a dominant vendor's settings decide what millions of children can access, its design works like regulation but answers to no one democratically. In November 2026 Google set Gemini and Search to "off by default" for K-12 accounts. Search now carries AI Overviews and AI Mode, so schools were really choosing between "Search with AI" and "no Search" [1]. Accountability must be shared, but it should rest most heavily on those who hold the power.

How product design became policy

• Governance vacuum: fewer than 10% of 450+ schools and universities had a GenAI policy [2]. Only 16% of countries guarantee data privacy in education by law [3].
• Bundling: AI built into an essential tool makes "opting out" an illusion, so the vendor's default becomes the rule [1].
• Continuing harm: a US lawsuit alleging that Gemini encouraged a teenager's suicidal thoughts [1] shows that a one-time switch cannot police what a chatbot says every day.

Who should be accountable

• Technology company, which owes the main duty of care: it designs the risk and profits from it. It should let schools switch AI off without losing Search, make outputs safe for children and submit to audits. With 89% of 163 EdTech products able to monitor children [3], default-off should be the minimum standard, not a favour.
• MeitY / Data Protection Board: the DPDP Rules, 2025 require verifiable parental consent and ban tracking, behavioural monitoring and targeted ads aimed at children, with penalties of up to ₹200 crore [4]. They should clarify that the education exemption [4] covers the school's teaching, not a vendor training its models on student data.
• Competition Commission of India: it should examine whether tying AI to a dominant Search product is abuse of dominance under the Competition Act, 2002.
• Ministry of Education and the states: they should publish GenAI guidelines. The OECD found that none of 18 jurisdictions had specific regulation of GenAI in education [5]. States can use their purchasing contracts to demand separate switches and a ban on training AI with student data.
• Schools and teachers: they should use AI selectively, for a clear teaching purpose. The OECD notes that general-purpose tools can improve task performance without producing real learning gains [6].

Vendor design ──► School default ──► Child's experience
  ▲ CCI / DPB       ▲ MoE / State rules    ▲ Teacher judgment

Fig.: Each level of decision needs its own accountable check.

Product design becomes public policy only when the state leaves a gap. Accountability should therefore follow power: companies are responsible for safety by design, regulators set enforceable limits, and educators decide how AI is used. The right to privacy (Puttaswamy, 2017), the right to education under Article 21A and SDG 4 all require classrooms to be governed by public rules, not by vendor defaults.