Misinformation and Deepfakes: Can We Still Trust What We See Online?

By Navyug News |

Author - Dr. Upananda

Legal and regulatory context checked on 3 October 2026

Imagine receiving a video call from someone who looks and sounds like your brother. He appears frightened. He needs money immediately, asks you not to call anyone else, and sends a payment link. You recognize the face. You recognize the voice. But the caller may be a stranger.

This is a hypothetical scenario, yet it captures a disturbing possibility: a familiar appearance can now be manufactured. CERT-In has warned about deepfakes impersonating relatives and executives to obtain money or sensitive information. The question facing internet users is no longer simply whether a picture has been edited. It is whether the person, event or statement depicted existed at all.

Trust online remains possible. But it must be built through evidence, context and verification.

A genuine video can still tell a false story

Misinformation generally means false information shared without necessarily intending deception. Disinformation involves deliberate deception. Deepfakes are AI-generated or manipulated images, audio and videos that imitate reality.

The distinctions matter because many misleading posts need no sophisticated technology. An old flood video can be presented as today’s disaster. A genuine speech can be cropped before a crucial qualification. A real photograph can acquire an invented location.

A clip can therefore be authentic while its caption is false. Verification must examine the claim surrounding the image, not just the image itself.

Nor is every synthetic image deceptive. Clearly labelled illustrations, accessible educational content and authorized creative work can serve legitimate purposes. Problems arise when fabrication is concealed, consent is ignored or invented material is presented as evidence.

The emotional trap behind the share button

A message claiming “the media will never show you this” invites you to become its distributor. A shocking clip rewards immediate outrage. A familiar face offers instant reassurance.

The dangerous moment is often the interval between feeling something and checking something.

Ask what the post wants you to do: transfer money, hate a community, distrust an election, buy a product or forward a warning. Urgency and secrecy deserve scrutiny, especially when accompanied by payment requests.

A trusted friend can unknowingly pass along misinformation. Their honesty does not establish the original source’s reliability. Thousands of shares may represent thousands of people relying on the same unverified upload.

Elections: when a fabricated clip can distort a choice

Imagine a candidate apparently withdrawing on the eve of polling, or an audio message claiming that voting has been postponed. These hypothetical examples show how misinformation can affect participation as well as political opinion.

During India’s 2024 general election, manipulated videos falsely portrayed actors Aamir Khan and Ranveer Singh making political statements. Reuters documented the videos and reported that both actors said they were fake. A celebrity’s recognisable voice and face had become vehicles for an endorsement they had not made.

For election information, check the Election Commission of India, the relevant Chief Electoral Officer and district election authorities. Confirm the constituency, date and original notice. A screenshot carrying a logo is not equivalent to a document published by the issuing authority.

What Indian laws actually cover

India’s response combines criminal law, election law and intermediary regulation. The relevant offence depends on the facts; “fake news” is not a universal criminal charge covering every incorrect statement.

The Information Technology Act, 2000: Section 66D concerns cheating by personation through a communication device or computer resource. A cloned voice used to impersonate a relative and fraudulently obtain money may attract it. Section 66C concerns dishonest or fraudulent use of another person’s electronic signature, password or unique identification feature; a copied face alone does not automatically satisfy its elements.

Sections 67 and 67A address publication or transmission of obscene and sexually explicit electronic material respectively. Section 67B provides protections concerning sexual material involving children. Non-consensual sexual deepfakes can raise these provisions, depending on the content and statutory requirements. Section 66E specifically concerns images of a person’s private area and its defined privacy conditions; it should not be described as applying automatically to every impersonation video.

The Bharatiya Nyaya Sanhita, 2023: Section 319 addresses cheating by personation. Section 336 concerns forgery, including qualifying false electronic records made with the required intent. A fabricated document is not automatically forgery merely because someone dislikes it; the legal ingredients must be established.

Section 353 addresses statements conducing to public mischief, including specified false information or rumours circulated electronically. It targets conduct meeting requirements concerning consequences such as public fear linked to offences, or intergroup enmity. It also contains a good-faith exception subject to its conditions.

Section 356 addresses defamation, including harmful imputations through visible representations, subject to statutory exceptions. A fabricated video falsely attributing misconduct may raise defamation issues. Legitimate criticism, factual reporting and protected expression require separate consideration.

Election law adds another layer

Section 123(4) of the Representation of the People Act, 1951 treats specified false factual statements about a candidate’s personal character, conduct, candidature or withdrawal as a corrupt practice. Its conditions include publication by the candidate or agent, or another person with the relevant consent, the specified belief about falsity, and likely prejudice to electoral prospects.

This is not a blanket rule making every inaccurate political opinion an electoral offence.

The Election Commission’s directions of 6 May 2024 warned political parties against deceptive deepfake audio and video. Parties were instructed to remove the specified misleading content within three hours of it being brought to their notice. That direction to political parties should not be confused with separate obligations imposed on platforms.

What changed for platforms in 2026?

The notified Information Technology Rules amendments, G.S.R. 120(E), took effect on 20 February 2026. They introduced provisions concerning “synthetically generated information”, with a defined scope and exclusions for certain routine, good-faith editing and other uses.

Rule 3(3) imposes specified safeguards on covered intermediaries concerning unlawful synthetic content, labelling and technically feasible provenance measures. Rule 4(1A) requires significant social media intermediaries to obtain user declarations and take proportionate technical measures to check them and label confirmed synthetic content.

The amendments also shortened the Rule 3(1)(d) removal period to three hours following qualifying legal notice. That does not mean every disputed post must disappear within three hours of an ordinary user complaint.

Labelling improves transparency but cannot replace verification. An absent label does not prove authenticity, and a synthetic label does not tell readers whether the accompanying factual claim is true.

Protecting truth without silencing debate

Article 19(1)(a) protects freedom of speech and expression, while Article 19(2) permits reasonable restrictions on specified grounds. Efforts to counter deception must respect that constitutional balance. An inconvenient investigation, unpopular opinion or clearly understood parody should not be casually equated with a fraudulent impersonation.

Platforms should explain moderation decisions and provide meaningful complaint and appeal routes. Authorities should identify the applicable legal provision and evidence. Readers should demand the same verification standard for claims they welcome and claims they reject. Otherwise, selective scepticism becomes another tool of manipulation.

A practical verification routine-

Pause before reacting. If a post produces immediate panic or outrage, delay sharing. Unsupported urgency is a warning signal.

Find the original. Search the exact wording, speaker and event. Look for the full recording, earlier uploads and reporting that identifies its source.

Check time and place. Compare visible signs, landmarks and event details. Reverse-image search a photograph or selected video frames to find previous appearances.

Compare independent evidence. Five websites repeating one anonymous post are not five independent confirmations. Seek sources that explain how they know.

Verify requests through another channel. For money or confidential information, call the person using a number you already know. Follow established workplace approval procedures. CERT-In recommends source checks, cross-referencing and callback procedures for sensitive transactions.

Treat detection tools as supporting evidence. Strange lip movements or lighting can justify further investigation, but their absence proves nothing. Researchers have documented limitations in deepfake detection. A detector score should not be treated as a final verdict.

If you are targeted, preserve and report

Save URLs, account handles, timestamps, messages and payment details. Keep relevant evidence securely without publicly redistributing humiliating or intimate material.

Report impersonation or harmful content through the platform’s reporting and grievance channels. Suspected criminal conduct can be reported to the police or the National Cyber Crime Reporting Portal. For cyber financial fraud, call 1930 promptly and contact your bank.

Newsrooms must carry their share of responsibility: verify originals, contact affected people, explain uncertainty and publish visible corrections. Reposting a false clip under a sensational headline can amplify its damage even when the article eventually debunks it.

The healthiest response to deepfakes is neither blind belief nor blanket disbelief. Real evidence should not be dismissed simply because fabrication is possible.

Before forwarding a dramatic clip, ask: Who published it? Where is the original? What independent evidence supports the claim? Those questions turn passive viewing into informed judgment. In a crowded digital world, trust is earned through checking—and sometimes protected by choosing not to share.