Off the guard rails: On the Grok case, explicit imagery

Beyond Editorial

Challenges Posed by Artificial Intelligence (AI)

• Design Challenges: Data bias, opaque algorithms, lack of explainability, weak safety guardrails, and large-scale amplification of harmful outputs (e.g., biased facial recognition systems misidentifying women and minorities).
• Ethical Challenges: Erosion of consent, misuse of personal data and identities, threats to dignity and autonomy, and unclear moral responsibility for harm (e.g., AI-generated deepfake pornography).
• Gender-Related Challenges: Disproportionate targeting of women through non-consensual synthetic imagery, online abuse, reinforcement of stereotypes, and hostile digital spaces (e.g., deepfake images of women journalists and politicians).
• Economic Challenges: Job displacement from automation, widening skill gaps, uneven growth benefits, and market concentration (e.g., AI-driven automation in BPOs and manufacturing).
• Cross-Jurisdictional Challenges: Borderless AI deployment, conflicts over applicable law, regulatory arbitrage, and enforcement difficulties (e.g., platforms headquartered abroad affecting Indian users).
• Strategic Challenges: Use of AI in cyberattacks, surveillance, autonomous systems, and information warfare (e.g., AI-enabled cyber intrusions and drone warfare).

India’s Approach to Artificial Intelligence (AI)

• Global leadership: At GPAI Summit 2023, hosted under India’s chairmanship in New Delhi, India positioned itself as a key voice in global AI governance.
• People-centric vision: PM Modi emphasised using AI for public welfare, with a special focus on ensuring that countries of the Global South can access and benefit from AI for inclusive development.
• Trust and safety: India underscored the need for a regulatory framework that ensures AI systems are safe, trusted, and reliable, while promoting international collaboration for long-term and scalable adoption.
• Regulatory philosophy: Rather than regulating AI at every stage of development, India supports platform-level guidelines, focusing on managing risks such as bias, misuse, and ethical concerns during model training and deployment.

Way Forward to Address Challenges Posed by AI

• Strengthening Design and Technical Safeguards: Mandate bias audits, explainability standards, and safety guardrails, with continuous testing to prevent harmful outputs (e.g., accuracy audits for facial recognition after documented misidentification of women and minorities).

• Embedding Ethical Governance: Establish clear norms on consent, data use, and identity protection, with enforceable liability for harm (e.g., criminalisation and takedown obligations for AI-generated deepfake pornography).

• Gender-Sensitive AI Frameworks: Require safeguards against non-consensual synthetic content, gender bias testing, and fast redress (e.g., platform-level filters and expedited complaint handling for deepfake abuse targeting women journalists).

• Managing Economic Transitions: Scale reskilling and promote human–AI collaboration while curbing concentration (e.g., national skilling initiatives aligned to automation impacts in BPOs and manufacturing).

• Addressing Cross-Jurisdictional Gaps: Enhance international cooperation and extraterritorial enforcement (e.g., compliance orders for foreign platforms affecting Indian users).

• Mitigating Strategic and Security Risks: Adopt national AI security doctrines and regulate high-risk uses (e.g., controls on AI-enabled cyber intrusions and autonomous drone applications).