Source: The Hindu

AI Misuse and the Rise of AI “Uplift”

• Anthropic’s September 2026 threat-intelligence report highlights a shift from AI being merely a content-generation tool to an “orchestration layer” capable of coordinating multiple stages of sophisticated operations.
• AI “uplift” refers to the enhancement of harmful capabilities through AI in terms of speed, scale and depth, reducing the expertise, manpower and time required.
• Anthropic documented AI misuse across seven broad categories involving state-sponsored actors and individuals.
• Bangladesh case:
• An AI-enabled disinformation operation combined content generation with automated video production, account creation and algorithmic scheduling.
• One individual reportedly operated 29 accounts and generated around 1,500 fabricated stories.
• Detected cases may underestimate the actual scale of misuse because they represent only activity identified by platforms.
• Existing safeguards also have limitations: Anthropic has acknowledged that its most capable models cannot guarantee that sophisticated actors will not obtain meaningful assistance in biological-weapons research.

Limits of Voluntary AI Restraint

• Anthropic CEO Dario Amodei has called for slowing frontier-AI development amid concerns over increasingly autonomous AI-agent “swarms” and their potential for large-scale disruption.
• A voluntary industry-wide slowdown faces structural constraints:
• Intense commercial competition among AI companies.
• Large private-capital investments in frontier AI.
• Geopolitical competition, particularly involving China.
• Difficulty in ensuring simultaneous compliance by all major developers.
• Voluntary restraint can also create potential conflicts of interest because firms advocating slower development may simultaneously face competitive incentives.
• Therefore, voluntary commitments alone cannot provide a durable governance mechanism.
• The policy focus needs to shift from attempting to slow technological progress to developing governance mechanisms capable of keeping pace with it.

Expanding Platform Accountability to Emerging AI Risks

• India’s Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021 demonstrate that legal responsibilities can be imposed on large digital platforms.
• Bangladesh’s AI-enabled disinformation case has particular relevance for India because of its large electorate, linguistic and social diversity, extensive digital penetration and frequent electoral activity.
• AI can make multilingual disinformation cheaper, faster and more scalable, while detection becomes increasingly difficult.
• Platforms therefore require enforceable duties to detect, disrupt and report coordinated AI-enabled influence operations.
• Emerging risks extend beyond disinformation:
• Model distillation — extracting or reproducing capabilities of another AI model.
• Capability extraction and fraudulent mass API access.
• AI-enabled surveillance and large-scale information analysis.
• Increasing AI-enabled surveillance raises concerns regarding excessive concentration of power and the Right to Privacy under Article 21 of the Constitution.
• Governance must therefore cover not only AI-generated content but also AI infrastructure, access, autonomy and downstream actions.

Proposed Guardrails for India

• Mandatory misuse reporting: AI platforms above a specified scale could be legally required to report detected misuse to CERT-In and a designated AI Safety Authority.
• Content provenance: AI-generated material, particularly in political and public-interest contexts, could require watermarking or other provenance mechanisms for traceability.
• Agentic AI regulation: Existing platform-accountability frameworks should explicitly cover AI systems capable of independently taking actions rather than merely generating information.
• Model-distillation safeguards: Systematic unauthorised model distillation and fraudulent mass API access should be addressed through specific legal provisions.
• Statutory AI safety institution: An independent authority could be empowered to:
• Compel relevant disclosures.
• Conduct safety and compliance audits.
• Investigate systemic AI risks.
• Impose proportionate restrictions where necessary.
• Such regulation should be risk-based and technology-neutral so that governance remains relevant as AI capabilities evolve.

India’s Opportunity: Innovation with Accountability

• India can develop an AI governance model that combines technological innovation with enforceable accountability.
• Experience with Digital Public Infrastructure (DPI), including UPI and Aadhaar, demonstrates India’s capacity to deploy and regulate technology at population scale.
• A comprehensive framework should integrate:
• Innovation and technological development.
• Transparency and accountability.
• Cybersecurity.
• Privacy and fundamental rights.
• Public safety and democratic institutional safeguards.
• The objective should be to prevent excessive concentration of AI capabilities and decision-making power among a small number of technology companies.
• With agentic and increasingly autonomous AI, frameworks designed primarily for conventional social-media platforms or generative content are becoming insufficient.
• India therefore needs governance that evolves alongside AI capability: development may accelerate, but accountability, safety and oversight must accelerate faster.