
The July 28 news is a commitment, not a new Instagram screen
Meta announced on July 28 that it is signing the EU AI Act Code of Practice on Transparency of AI-Generated Content. The company framed the move as part of its work on practical, interoperable ways to identify increasingly realistic synthetic media.
That does not mean a brand-new label reached every Instagram account on July 28. Meta’s post does not provide a UI rollout date, and the company has used AI info labels across Instagram, Facebook, and Threads since 2024.
AI info is a clue, not a final authenticity ruling
Meta says its labels can be triggered by industry-shared indicators or by a creator disclosing that AI was used. Content detected as AI-generated can receive a visible label, while information about some AI-edited content may appear inside the post menu.
That distinction matters. A label can describe an image created from scratch, but it can also reflect an editing workflow. Open the label and read the reason before assuming that every part of the post was fabricated.

No label does not prove that a photo came straight from a camera
The system depends in part on technical signals, platform detection, and creator disclosure. Those signals may be missing, removed during processing, or not recognized. Meta does not claim that every synthetic image will always be labeled.
The reverse is also important: a label does not automatically make every factual claim in a post false. A real photograph may have used a generative fill or another AI-assisted edit. Provenance and truthfulness are related questions, but they are not the same question.
Raise the verification bar for consequential claims
For elections, disasters, public figures, breaking news, or allegations about a real person, do not stop at a screenshot. Find the earliest upload you can, inspect the account’s history, and compare the post with official statements and independent images or video from the same event.
Visual oddities can justify a closer look, but they are not conclusive evidence. Use them to start a source check rather than to declare a post real or fake from appearance alone.
Creators should pair disclosure with good source records
If you publish realistic AI-generated or AI-altered media, use the platform’s disclosure control and explain the production method when context matters. Keep the camera original, pre-edit export, tool name, and creation date so you can document the work later.
The European Commission describes separate duties for AI interaction notices, machine-readable marks, deepfakes, and some public-interest text. The Article 50 rules begin applying on August 2, with scope rules, exceptions, and transitional arrangements. Businesses serving people in Europe should read the current official guidance instead of treating this reader guide as legal advice.

One last check
Five checks before you reshare
- 01
Open the AI info label or the post’s three-dot menu and read the explanation.
- 02
Do not treat the absence of a label as proof that a photo is authentic.
- 03
For important people, places, or events, look for the earliest source and an official account.
- 04
Compare the date, account identity, and independent photos or video before sharing.
- 05
If you publish realistic AI media, use the disclosure tool and retain the original files.
Sources
Sources used for this article
- Meta — Signing the AI-generated content transparency code
- Meta — How AI info labels use detected signals and disclosure
- European Commission — Article 50 transparency guidelines
- European Commission — AI transparency quick facts
- The Japan Times / AFP-Jiji — Report on the EU labeling rules
Sources checked July 28, 2026. Meta’s same-day announcement concerns signing the EU transparency code; the AI info labeling system predates it. EU transparency requirements begin applying on August 2, subject to scope, exceptions, and some transitional arrangements. The article images are editorial concepts, not real Instagram screens or evidence that any depicted media is authentic.
Was this guide useful?
This browser can leave one heart per article.
