AI-Enabled Image Abuse and Deepfake Sextortion: Safety-by-Design Lessons for Women in India

CASE STUDY 19 / 200

India — emerging digital harms and cross-cutting technology policy

Independent desk-research draft; illustrative synthesis, not a forensic investigation

Abstract

This case study considers how generative AI and image-manipulation tools can intensify gendered online abuse through fabricated intimate imagery, impersonation, sextortion and coordinated harassment. It draws on a 2025 Indian civil-society research report mapping AI-facilitated gendered harm, a 2026 parliamentary committee summary on cybercrime and cyber safety of women, and official government information about deepfake-related legal and regulatory measures. It does not reproduce graphic material or infer that every manipulated image is AI-generated.

Keywords

deepfakes; AI-enabled abuse; image-based sexual abuse; sextortion; privacy; safety-by-design

1. Introduction and Research Question

What safeguards should a women’s safety app implement when abuse includes synthetic media, impersonation or threats to distribute manipulated images?

2. Methodology and Evidence Selection

Desk review of a 2025 report by RATI Foundation, a Parliamentary Standing Committee report summary published by PRS Legislative Research in March 2026, and official government material. The report’s case examples are third-party documented cases; this case study does not independently verify the underlying incidents or conduct forensic analysis.

3. Case and Context

Synthetic or manipulated images can be used to threaten, shame, extort or impersonate a person. A victim may not know where content has been shared, whether the image is synthetic, or which reporting route is most appropriate. Harm can persist even if one copy is removed. Automated detection may produce false positives and cannot reliably determine consent, context or legal guilt on its own.

4. Evidence-Based Findings

• The RATI Foundation’s 2025 report, Make It Real: Mapping AI-Facilitated Gendered Harm, documents reported examples involving suspected AI-enabled image abuse, threats, doxxing and sextortion. Its cases are illustrative evidence, not a population prevalence estimate.
• A PRS India summary of the Parliamentary Standing Committee’s report on cyber crimes and cyber safety of women, submitted on 23 March 2026, discusses cyberstalking, sextortion, deepfakes, online harassment, investigative capacity and platform accountability.
• Government communications in 2026 describe existing legal provisions and regulatory measures relevant to aspects of deepfake misuse. The applicable offence and remedy depend on the specific conduct, content, intent and facts.
• Detection tools, hash matching, takedown requests and law-enforcement reporting serve different purposes. None guarantees complete removal or prevents re-uploading.

5. Discussion

The safest product response is practical and non-sensational: help the affected person preserve a record of threats, reach trusted support, find official reporting options and understand likely next steps. Avoid requiring the user to upload the intimate image to VETRIPEN. If evidence capture is offered, minimise copies, use secure storage and explain deletion limits. Never label a person as an offender solely on an automated model’s output.

6. Legal and Policy Context

This report is not a legal opinion. India’s Information Technology Act, 2000 and other laws may apply to identity theft, impersonation, privacy violations and publication/transmission of certain content, but legal classification depends on the facts and current law. Verify the latest official text and seek qualified legal assistance where needed.

7. Potential VETRIPEN Design Implications

The following are product-design hypotheses for evaluation, not proven interventions:

• Provide a guided response checklist for threats involving manipulated images, with optional steps and clear warnings about evidence handling.
• Link to the official National Cyber Crime Reporting Portal and verified platform-reporting pages.
• Support safe documentation of account handles, timestamps, URLs and threat messages without requiring the app to ingest intimate imagery.
• Keep all detection or risk scores clearly labelled as uncertain; never present them as a legal finding.
• Do not send automatic notifications to relatives, employers or public contacts; require user-controlled consent and disclose exceptions.

8. Suggested Evaluation Metrics

• Time and task success for finding a verified reporting route in usability tests.
• User comprehension of limitations of detection, takedown and evidence preservation.
• Security assessment of storage, permissions, retention and data export.
• Number of false or misleading automated classifications in any evaluated prototype; independent review before release.

9. Limitations and Verification Status

This is a desk-based synthesis and does not establish how common AI-enabled image abuse is in India. The RATI report’s individual cases are not independently re-investigated here. Parliamentary summaries are not substitutes for the complete report, and legal rules may change. Do not use the case examples as if they were VETRIPEN users or original interviews.

10. Conclusion

AI-enabled abuse requires survivor-centred response, privacy protection, realistic claims and dependable referral routes. VETRIPEN should help people navigate support and document threats safely, not claim to identify guilt or guarantee removal of synthetic content.

11. References and Source Notes

Sources were identified through desk research. Their presence below does not imply that every claim has been independently audited. Review the original documents and record access dates before external publication.

1. RATI Foundation. Make It Real: Mapping AI-Facilitated Gendered Harm. 2025. https://ratifoundation.org/wp-content/uploads/2025/11/RATIFoundation_ResearchReport_Make-It-Real.pdf
2. PRS Legislative Research. Cyber Crimes and Cyber Security of Women: Standing Committee Report Summary. 23 March 2026. https://prsindia.org/policy/report-summaries/cyber-crimes-and-cyber-security-of-women
3. Press Information Bureau, Ministry of Electronics and Information Technology. Government Strengthens Regulatory Framework to Address AI-Generated Deepfakes. 6 August 2026. https://www.pib.gov.in/PressReleasePage.aspx?PRID=2295500&lang=2&reg=48
4. Government of India. National Cyber Crime Reporting Portal. https://cybercrime.gov.in/

12. Declarations

Research type: secondary desk research. No new human-participant data were collected for this report.

Peer review: not peer reviewed.

Conflicts/funding: prepared as part of Digital Catalyzt’s independent research and product-design programme; no external funding is declared in this draft.

AI assistance: AI tools assisted with drafting and organisation. Human source verification and editorial review are required before publication.

DISCLAIMER — INDEPENDENT RESEARCH DOCUMENT. This document has been prepared privately by Digital Catalyzt as part of an independent research and product-design programme related to women’s safety and the proposed VETRIPEN service. It is provided for research, education, discussion and preliminary product-planning purposes only. Unless expressly stated and documented, it has not undergone independent peer review and is not a journal publication, government statement, legal opinion, medical or psychological assessment, or formal investigation. Claims are limited to the sources and scope stated; sources may contain limitations, errors, incomplete data or later revisions. Allegations must not be presented as established findings unless an authoritative process has established them. Hypothetical examples must not be represented as real events or interviews. Legal and policy discussion is general information, not legal advice; verify current rules with qualified professionals or responsible authorities. Proposed VETRIPEN features are design hypotheses unless independently evaluated. This document does not guarantee prevention of harm, emergency response, message delivery, confidentiality in every circumstance or any particular outcome. Readers should assess the cited evidence, limitations and local context independently.

← Back to all research