Source, label, date, context
Who posted it first?
Most viral images and clips have been reposted many times. Try to find the earliest version and the person or organisation behind it. A reverse image search (Google Lens, Bing Visual Search or TinEye) shows where else a picture has appeared. For video, search a screenshot of a key frame. If the trail leads to a known news agency or the person's verified account, that is real evidence.
Is there a label or credential?
Many platforms now label AI-generated content, and some cameras and tools attach Content Credentials, a standard record of how media was made and edited, which you can inspect at contentcredentials.org/verify. A credible label is useful evidence. But the absence of a label proves nothing, because labels are often stripped when media is downloaded and re-uploaded.
When, and does the context fit?
Old photos are often recycled as "breaking news": a reverse search may show the same picture from a different event years ago. Check that the weather, location, clothing, signs and other coverage fit the claim. If something dramatic really happened, several independent news organisations will usually be reporting it within hours.
"Not enough information" is fine
Often the honest conclusion is that you can't tell yet. The right move then is simply not to share it, or to share only with a clear note and a trusted source once one exists. It is worth remembering the upside too: the same tools restore old family photos, power film effects and help people with disabilities. The problem is not generation; it is passing something off as what it isn't.
Pick an image you saw shared today. Could you find where it first appeared?