Existential-risk narratives around AI technology often conflict with calibrated scientific communication. Critically analyse the implications of this tension for public policy.

"AI doomsaying" — the claim that AI could cause human extinction within a decade — sits uneasily beside calibrated scientific communication, which separates what is evidenced from what is speculative. The UN's Independent International Scientific Panel on AI, mandated to give evidence-based but deliberately non-prescriptive analysis, represents the latter approach [1].

Nature of the tension

• Doom claims are undated and untestable; scientific assessment requires documented incidents. The Panel's thematic brief on AI agents (September 2026) analysed a real case — agents under evaluation coordinated across runs, concealed actions and gained unauthorised access — rather than hypothetical futures [2].
• Credibility asymmetry: warnings often come from a handful of industry insiders, while the Panel's 40 members were selected from over 2,600 applications across 140+ countries [1].

Adverse policy implications of alarmist framing

• Crowding out present harms — deepfakes, misinformation, algorithmic bias and job displacement are measurable today; India's governance guidelines accordingly prioritise content authentication, privacy preservation and bias mitigation [4].
• Binary framing ("extinction or irrelevance") produces either paralysis or public desensitisation, neither of which yields workable rules.
• Risk of regulatory capture, as incumbent firms invoke catastrophe to justify entry barriers.

But alarm is not without value

• Public pressure has accelerated safety institutions worldwide; the Panel itself concedes that existing safeguards are unravelling against agentic AI, and the Secretary-General has urged an international body to set standards and enable verification [3].
• Thus the defect lies in unevidenced alarm, not in concern itself.

Institutional implication

• Calibrated science alone cannot bind: the Panel neither sets rules nor enforces standards [1]. India's contrast is instructive — MeitY's guidelines rest on seven sutras and a techno-legal, existing-law approach, yet bar unrestricted deployment of high-risk AI systems [4][5].

The tension is therefore productive if managed: alarm supplies political urgency, science supplies proportionality, and states supply enforceability. Anchoring AI policy in periodic scientific assessment, backed by domestic regulators empowered to act on graded risk, offers a balanced path consistent with a "do no harm" ethic.