Image Manipulation and Context Hijacking

The phrase 'seeing is believing' captures a deep human intuition: visual evidence feels more trustworthy than verbal testimony, more immediate than a written argument, more visceral than a statistic. That intuition has always been somewhat misplaced — photography has never been a neutral record of reality — but in the contemporary information environment it has become genuinely dangerous.

Research on visual misinformation distinguishes two fundamentally different mechanisms by which images mislead (Wardle, 2017). The first is image manipulation: the pixels themselves have been altered. A face is swapped. A crowd is expanded. A sky darkens to suggest a more dramatic scene than existed. The second, and by far the more prevalent, is context hijacking: the image is entirely authentic, but it is being circulated to illustrate an event, claim, or location with which it has no connection.

The distinction matters practically, because they call for different verification approaches and carry different implications for trust.

Why context hijacking dominates

Context hijacking is more common than pixel-level manipulation for a straightforward reason: it is much easier to do. Reusing a photograph requires no technical skill — only the decision to circulate it alongside a misleading caption. A photograph of flooding in one country circulates as 'evidence' of flooding in another. A protest photograph from five years ago resurfaces as documentation of a recent event. A dramatic image of smoke and flame from an industrial accident in one city illustrates a terrorist attack in a different continent.

First Draft News, a coalition organisation that monitored and researched visual misinformation (note: First Draft was an organisation that monitored visual misinformation; note that it ceased operations in 2023 — its research archive remains available and is cited here for its methodological contributions) since 2015, has documented this pattern extensively. Their training resources on verifying online information note that the majority of harmful visual content encountered by journalists and fact-checkers involves authentic images stripped of their original context rather than technically falsified ones (First Draft, 2021).

What pixel-level manipulation actually looks like

Manipulation of image content — as opposed to context — ranges from crude to sophisticated. At the crude end: objects removed or inserted using basic editing tools, leaving visible artefacts around edges. Photographs of crowds inflated by copy-pasting groups of people. At the sophisticated end: AI-generated faces composited into real scenes, or lighting and shadow adjusted so carefully that inconsistencies are invisible to casual inspection.

The important point is that even sophisticated pixel-level manipulation is detectable by trained analysts using the right tools. What is often not detectable is the context — when, where, and why an authentic image was actually taken.¹

Footnotes#

1. The challenge is asymmetric: detecting that an image has been manipulated is a technical question that can often be resolved. Establishing the original context of an authentic image is a historical and investigative question that requires different methods entirely — including geolocation, open-source research, and cross-referencing with archived sources. ↩