Self-certification by private companies in safety-critical technologies raises questions of accountability and public trust. Discuss in the context of autonomous vehicle deployment.

Self-certification lets a manufacturer itself declare that its product conforms to mandatory safety standards, with the regulator verifying only afterwards. Tesla's steering-less Cybercab robotaxi, commercially launched in Austin in September 2026 and placed under a federal NHTSA probe within hours, illustrates both the appeal and the risks of this model [1].

Why self-certification is used

• Regulatory lag: standards written for human-driven cars — steering wheels, pedals, mirrors — cannot literally apply to a purpose-built driverless two-seater, so firms declare such clauses inapplicable [1].
• Speed and cost: it suits rapid technology cycles and avoids case-by-case approval delays.
• Capacity asymmetry: regulators often lack the proprietary sensor and software data needed to test autonomous systems independently.

Accountability concerns

• The firm becomes judge in its own cause — Tesla self-certified rather than seeking a formal exemption, the very issue now under scrutiny [1].
• Enforcement turns ex-post: scrutiny began only after a fleet of roughly 1,000 vehicles was already carrying passengers on public roads [1].
• The exemption route offers a contrast: NHTSA's first-ever demonstration exemption for an American-built driverless vehicle came with volume caps and continuing operational oversight [2].
• Liability is diffused — with no driver to hold responsible, fault shifts to untested product-liability and algorithmic-accountability questions.

Public trust

• Passengers with no manual override cannot intervene in an emergency, raising informed consent issues; reported erratic driving during the pilot deepened unease [1].
• Opaque, proprietary algorithms mean trust rests on disclosure, independent audit and transparent incident reporting.

Indian context

• The Motor Vehicles Act, 1988 and CMVR prescribe prior type approval by designated test agencies rather than self-declaration — a safer starting point, though untuned to autonomy [3].
• NITI Aayog's National Strategy for AI flags smart mobility as a priority needing safety and ethics frameworks [4].

Self-certification is not inherently illegitimate; unsupervised self-certification is. A graduated model — conditional exemptions, mandatory data and crash disclosure, third-party audit and clear liability rules — can let innovation proceed while keeping the state's precautionary duty intact. Public safety, ultimately, is a non-delegable state responsibility.