"Safe harbour under Section 79 is conditional, not absolute." Discuss with reference to morphed content and platform accountability.

Section 79 of the IT Act, 2000 exempts intermediaries from liability for third-party content, but only so long as they observe due diligence prescribed by the Government. The Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021 convert that proviso into an enforceable code, making immunity a conditional privilege earned by conduct, not a blanket shield.

How the law builds in conditionality

• Rules 3 and 4 impose due-diligence duties — publishing rules, grievance officers, compliance reports; failure forfeits the Section 79 exemption [1].
• Section 79(3)(b): immunity ends once the intermediary has "actual knowledge" and does not act. Shreya Singhal (2015) confined such knowledge to a court or government order, protecting speech but weakening a private victim's complaint [2].
• Significant social media intermediaries must additionally deploy automated tools and enable traceability [1].

Morphed content as the sharpest test

• Complaints about content that is sexual, or that impersonates including by morphing, must be acted on within 24 hours, and may be filed by the victim or on her behalf [1].
• MeitY's SOP on Non-Consensual Intimate Imagery (NCII) operationalises this through the I4C hash bank and DoT-ISP blocking [3].
• In September 2026 the Delhi High Court directed Meta to remove a morphed image of a woman circulated with the Prime Minister, and pressed the police on responsiveness — judicial enforcement of a duty the platform already owed [4].

Why accountability remains incomplete

• Automated matching covers only content identical to what was removed; a cropped or re-morphed file escapes, restarting the clock [1].
• Non-compliance carries no direct penalty — only eventual loss of immunity — so victims bear litigation costs [1][4].
• Conversely, fear of losing safe harbour invites over-removal; hence the 2025 amendment to Rule 3(1)(d), restricting takedown orders to senior officers with monthly review [5], and the 2026 rules on synthetically generated information requiring labelling of deepfakes [6].

Safe harbour is thus a bargain: immunity in exchange for diligence. Strengthening it requires near-identical hash matching for intimate imagery, victim-facing timelines under the NCII SOP, and active Grievance Appellate Committees — so that dignity under Article 21 is secured without chilling legitimate speech.