Assisted-by Tag

An assisted-by tag is a small marker that appears directly on a post, an image, or an article. It signals: this content was created or edited, wholly or in part, with an AI program. The word “tag” simply means a label here — comparable to the “sponsored” notice under an advertisement. The difference is that the assisted-by tag doesn’t indicate payment, but rather the use of software. Some platforms set the label automatically, while others require the user to set it themselves.

Trust as a scarce resource in the age of generated content

Modern AI programs can produce texts, photos, and videos in seconds that are barely distinguishable from human-made content. This creates a problem: someone reading a news article or looking at a photo often no longer knows whether a human or an algorithm is behind it. Trust in digital content suffers when this question remains permanently unanswered.

This is exactly where the assisted-by tag comes in. It aims to make this uncertainty visible without fundamentally condemning the use of AI. The idea is similar to mandatory labeling for food additives: you don’t have to avoid them, but you have the right to know they’re in there. For platforms, this is also a legal safeguard — in the EU, pressure is growing to make AI-generated content identifiable, for instance through the AI Act, the European Union’s new set of AI regulations.

Who sets the tag — and how reliable is it?

There are two basic approaches. In the first, the platform itself detects whether content is AI-generated — for example, through watermarks, meaning invisible markers that AI tools embed directly into images. In the second approach, users must indicate themselves when uploading whether they used AI. Platforms like Meta rely on a combination of both methods.

The weakest point is self-disclosure. Anyone who intends to deceive simply doesn’t check the box. Automatic detection is more reliable, but still not perfect: it sometimes triggers on ordinary photos or misses subtly edited content. The technology behind detection — such as metadata analysis or AI detectors — is currently developing rapidly, but is not yet a reliable cure-all.

There’s also the question of gradation. Is a photo “AI-assisted” if someone only blurred the background with an AI tool? Or only once the entire image has been generated? Most platforms have not yet drawn a sharp line here — which means the label is sometimes applied too generously and sometimes too rarely.

Where the tag is already appearing in practice

Meta introduced the assisted-by tag on Instagram and Facebook in 2024. Anyone uploading an AI-edited image there is supposed to label it — otherwise sanctions such as reduced reach may follow. LinkedIn is following suit and testing similar labels for posts written with the help of AI writing assistants. YouTube is also experimenting with labels for AI-generated videos, especially for realistic deepfakes — that is, deceptively authentic-looking fake video recordings of people.

The term also appears in journalism. Some news outlets label articles where AI helped with research or wording. The discussion there is lively: critics see the tag as devaluing the text, while proponents see it as a sign of honesty. In both cases, the assisted-by tag becomes a signal that readers increasingly interpret actively — it is no longer just a technical detail, but part of a piece of content’s credibility.