The screenshot proves they posted it: a circulating image of a public figure's social media post is reliable evidence that the post was made
By The Conspiratory EditorsSeptember 1, 2026
Where the evidence lands: ContradictedThat a screenshot or image of a social media post attributed to a politician, celebrity, or institution can be treated as evidence that the post was actually made, and shared or argued from on that basis.
Contradicted. This rates a habit, not any single screenshot: treating an image of a post as evidence the post exists. The volume now defeats it. On 31 August 2026 alone, Lead Stories ruled on four claimed Truth Social posts by the US president, finding three counterfeit and one genuine, and the genuine one carried an AI-generated video. The day before, Snopes had published something stranger than a debunk: a compilation of two dozen posts that are real, assembled because so many circulating screenshots are not. That inversion is the finding. When a fact-checker has to catalogue the authentic items, the ambient assumption has already flipped. Nothing here concerns the content or merits of any real post, and no claim is made about anyone's conduct; what is rated is the evidentiary weight of a screenshot, which on this record is close to none.
Believed by: Nearly everyone, across the political spectrum, and usually in the direction that flatters what they already think. A screenshot asks to be believed because it looks like a record rather than an assertion, and the people sharing it are rarely the people who made it
A second checker has now rated the forged mess hall post, and it found that the mockery carrying it was wrong about the picture. Snopes ruled on 2 September that the president did not post the image of a sailor eating a burger under a caption about the military eating the best, and Lead Stories had already traced the screenshot to a 4chan post of 30 August. What Snopes adds is a detail nobody sharing it seems to have checked. The line that gave the post its reach, that the troops are so well fed they have grown extra fingers, points at a sixth finger which is not visible in the image at all. What is visible is a hand with five non-thumb fingers and no thumb. The crowd reached the right verdict, fabricated and machine-made, by pointing at something that was not there. That is worth holding on to, because it cuts against the reassuring idea that people are getting better at this: an audience can be correct about a forgery while reasoning from a feature it has invented, and a habit of detecting AI by counting fingers will eventually meet an image with the right number. source →
The full story
Four rulings, one account, six hours
On 31 August 2026, Lead Stories published four separate adjudications of claimed Truth Social posts by the US president. Three were counterfeit: a mess hall photograph captioned about the military eating the best, itself AI-generated; an announcement of a Melania Canal to bypass the Welland Canal; and a threat to close New York's 9/11 memorial. Snopes covered the last of those the same day.
The fourth was real. A post about Kharg Island, published on the official account the previous day, carried an AI-generated video and a caption about the island being blown to smithereens. Lead Stories confirmed the post and recorded that no reporting indicates the video or such a strike is real.
So the day produced three fabrications and one authentic item, and the authentic item was the one containing synthetic footage. Any reader sorting these by how they looked would have got the most important one backwards.
When the checkers start listing what is real
The day before, Snopes published something that is not a debunk at all. Under the heading of two dozen things the president actually posted, it compiled genuine items from its archives, explaining that the newsroom often labels alleged posts as fake and traces them to people with digital editing skills.
Consider what that publication implies. A fact-checking organisation normally answers the question is this fake, one item at a time. Assembling a register of authentic posts answers a different question, which of these are real, and a newsroom only does that when the default assumption has flipped and readers can no longer take any screenshot at face value.
It is also an admission about scale. Checking individually does not keep up, so the response is to build a reference list instead. That works for one exceptionally prominent account. It does not obviously work for anybody else.
The test everyone uses, and why it broke
Most people sort these with a single question: would this person actually post that? It is a reasonable instinct and it has quietly stopped working, for a reason visible in the Snopes list itself.
Among the two dozen verified posts is an AI-generated image of the president as the Pope. The confirmed Kharg Island post carried AI video of a military strike. The collection also includes, in the outlet's description, racist imagery. Whatever one makes of any of that, it settles the evidentiary point: an account that genuinely publishes surreal and synthetic material cannot be assessed by whether an item seems too strange to be genuine.
The collapse runs both ways, which is what makes it expensive. A fabricated post is not exposed by seeming outlandish, and a real one is not defended by seeming ordinary. The plausibility test returns noise in both directions, and it is the only test most readers have.
The joke that was built on a hoax
The Melania Canal post is worth separating out, because the standard advice about satire fails on it in an instructive way.
Lead Stories traced it to a satire account, which sounds like the end of the matter. It is not, because of what the joke assumed. A bypass canal is only funny if there is a blockade to bypass, and the blockade was invented the previous day by a different fabrication: a claimed Fox News report that Canada was stopping vessels through the Welland Canal over the renaming of Lake Ontario. Fox News reported no such thing.
Underneath both sits something real, an executive order directing that Lake Ontario be styled Lake America in government publications during a dispute with Canada. So the ladder runs from a genuine order to a non-satirical hoax to a satirical post resting on that hoax. Checking whether the source is a satire account resolves the top step and leaves the middle one standing, and the middle one never announced itself as a joke.
Why good-faith sharing does most of the work
It is tempting to imagine a counterfeit post as the work of someone trying to deceive, passed along by people who half-know. Mostly that is not the shape of it. Whoever built the artifact is one person; the thousands who spread it are ordinary readers who found it convincing, which is why the material rarely carries any tell of bad faith.
A screenshot is persuasive before its content is read, because it arrives in the grammar of a record. There is a timestamp, an interface, a name, an engagement count. Those are the visual signatures of a thing that happened, and they are also the easiest elements in the entire artifact to fabricate, since they are just pixels arranged in a familiar pattern.
Verification, meanwhile, is genuinely awkward rather than merely tedious. Establishing whether a post exists means searching the account on the platform, which many people cannot do without an account of their own, and a deleted post leaves a hole that looks identical to a post that never existed. The effort is real and asymmetric: seconds to share, minutes to check, and sometimes no definitive answer at the end.
There is only one question, and it is not about the image
Nothing in the visual sorts these. Not the interface, not the plausibility, not the presence or absence of AI artefacts, since on this record the synthetic content sat inside the authentic post.
What sorted them, every time, was whether the post exists on the account. That is what Lead Stories established in all four rulings, and what Snopes built its register around. It is a question about the world rather than about the picture, and it is the only one worth asking.
The awkward part is that this test is easy for a newsroom and needlessly hard for a reader. Platforms could make any post resolvable at a permanent public address, readable without an account, which would reduce most of this to a click. That they largely do not is a design decision, not a technical constraint, and it leaves the burden with the people least equipped to carry it. In the meantime the honest position on any screenshot is the uncomfortable one: it is not evidence of anything until somebody checks the account, and most of the time nobody has.
What's still unexplained
- Whether platforms could make a post's existence cheaply checkable by outsiders. A permanent public URL for every post, resolvable without an account, would turn most of these into a one-click question, and its absence is a design choice rather than a technical limit.
- How to handle deletion honestly. A genuine post that is later removed leaves the same evidentiary hole as a post that never existed, and no method here distinguishes the two after the fact.
- Whether cataloguing authentic posts, as Snopes has begun doing, scales at all, or whether it works only for the handful of accounts prominent enough to justify the labour.
- What happens to this file's method when the account holder disputes an authentic post. Nothing in the record here turns on that yet, but a form of evidence this weak invites it.
Point by point
The claim: He would never post something that absurd, so the screenshot must be fake.
What the record shows: This test has stopped working, and the Snopes collection is the demonstration. Among two dozen verified posts is an AI-generated image of the president as the Pope. The Kharg Island post, confirmed authentic, carried AI video of a military strike. An account that genuinely publishes surreal and synthetic material cannot be defended by the argument that a surreal item must be invented, and the same collapse runs the other way: absurdity is no longer evidence either.
The claim: It looks exactly like the platform's interface, so it is real.
What the record shows: The interface is the easiest part to reproduce. Fonts, spacing, timestamps, verification badges and engagement counts are all just pixels, and reproducing them requires no special access. Every counterfeit in this file's record presumably looked correct, since an incorrect-looking one would not have travelled far enough to be checked.
The claim: A fact-checker would have flagged it by now if it were fabricated.
What the record shows: They are flagging them, at a rate that is itself the problem. Four rulings on one account in one day is not a backlog being cleared, it is a sample of an ongoing stream, and the corrections reach a fraction of the audience the originals do. Snopes assembling a list of authentic posts is the response of a newsroom that cannot check them one at a time.
The claim: The satire ones are obvious, and satire accounts label themselves.
What the record shows: The Melania Canal post did originate on a satire account, and that label did not survive the journey. Worse, the joke was built on a fabrication rather than on reality: it presupposed a Canadian blockade of the Welland Canal that a different, non-satirical hoax had invented the day before. Checking for a satire origin catches the last step in that chain and nothing behind it.
The claim: This only matters for politics, where people expect exaggeration anyway.
What the record shows: The same technique manufactures institutional records generally. The quotation attributed to the president's mother, unfounded since at least 2019, circulates with images made to look like newspaper clippings, and no real newspaper carried it. A clipping implies an archive, a masthead and a date; none exists, and the implication does its work at thumbnail size.
Timeline
- 2026-08-25Lead Stories rates as false a claim that the US president's Truth Social account posted an image of him walking on water with the caption Pastor, President, Protector. The image is not from the account.
- 2026-08-30Snopes publishes a collection of two dozen posts the president genuinely made on Truth Social, compiled from its archives. The framing is explicit: the newsroom often labels alleged posts as fake, tracing them to people with digital editing skills, so it has assembled the real ones. The set includes an AI-generated image of himself as the Pope, holiday messages, and material the outlet describes as racist imagery.
- 2026-08-30A post about Kharg Island appears on the president's official account, carrying an AI-generated video and a caption about the island being blown to smithereens.
- 2026-08-31Lead Stories rules on four claimed posts within six hours. A mess hall picture captioned about the military eating the best is fake, and the picture itself is AI-generated.
- 2026-08-31An announcement of a Melania Canal to bypass the Welland Canal is fake, and traced to a satire account. It depends on a premise invented the previous day by a different fabrication, a claimed Fox News report that Canada was blocking ships over the renaming of Lake Ontario.
- 2026-08-31A threat to close New York's 9/11 memorial over not being allowed to speak is fake. Snopes covers the same claim the same day.
- 2026-08-31Lead Stories confirms the Kharg Island post is authentic, published on the official account the day before, and records that no reporting indicates the video or such a strike is real. Three counterfeits and one genuine article in a single day, with the genuine one carrying the synthetic footage.
- 2026-09-01Snopes revisits a quotation attributed to the president's mother, unfounded and circulating since at least 2019, some versions carrying images built to resemble newspaper clippings. The same manufacture-the-record technique, applied to an older medium.
Contradicted. This rates a habit, not any single screenshot: treating an image of a post as evidence the post exists. The volume now defeats it. On 31 August 2026 alone, Lead Stories ruled on four claimed Truth Social posts by the US president, finding three counterfeit and one genuine, and the genuine one carried an AI-generated video. The day before, Snopes had published something stranger than a debunk: a compilation of two dozen posts that are real, assembled because so many circulating screenshots are not. That inversion is the finding. When a fact-checker has to catalogue the authentic items, the ambient assumption has already flipped. Nothing here concerns the content or merits of any real post, and no claim is made about anyone's conduct; what is rated is the evidentiary weight of a screenshot, which on this record is close to none.
Reviewed by The Conspiratory Editors · Last reviewed September 2, 2026 · How we rate
Common questions
Is Did they really post that? true?
Contradicted. This rates a habit, not any single screenshot: treating an image of a post as evidence the post exists. The volume now defeats it. On 31 August 2026 alone, Lead Stories ruled on four claimed Truth Social posts by the US president, finding three counterfeit and one genuine, and the genuine one carried an AI-generated video. The day before, Snopes had published something stranger than a debunk: a compilation of two dozen posts that are real, assembled because so many circulating screenshots are not. That inversion is the finding. When a fact-checker has to catalogue the authentic items, the ambient assumption has already flipped. Nothing here concerns the content or merits of any real post, and no claim is made about anyone's conduct; what is rated is the evidentiary weight of a screenshot, which on this record is close to none.
What is Did they really post that??
A screenshot of a post looks like documentation. It is not; it is a picture, and pictures of text are trivial to make. The week of 25 to 31 August 2026 showed what that now costs. Lead Stories ruled that the president's account had not posted an image of him walking on water captioned Pastor, President, Protector; had not posted a mess h…
What does the evidence show?
This test has stopped working, and the Snopes collection is the demonstration. Among two dozen verified posts is an AI-generated image of the president as the Pope. The Kharg Island post, confirmed authentic, carried AI video of a military strike. An account that genuinely publishes surreal and synthetic material cannot be defended by th…
Why do people believe it?
A screenshot presents itself as a record rather than a claim. It has the grammar of documentation: a timestamp, an interface, a name, and that grammar is persuasive before any content is read.
What is still unresolved?
Whether platforms could make a post's existence cheaply checkable by outsiders. A permanent public URL for every post, resolvable without an account, would turn most of these into a one-click question, and its absence is a design choice rather than a technical limit.
Sources
- 1.24 things Trump actually posted on Truth Social, Snopes (2026)
- 2.Fact Check: Trump DID Make Post On Truth Social About Kharg Island Using An AI-Generated Video, Lead Stories (2026)
- 3.Fact Check: Trump Did NOT Post 'Our Military Eats The Best' With AI-Generated Mess Hall Picture, Lead Stories (2026)
- 4.Fact Check: FAKE Trump Post Announces 'Melania Canal' To Bypass Welland Canal, It Originated From A Satire Account, Lead Stories (2026)
- 5.Fact Check: Trump Did NOT Post Threat To Close New York's 9/11 Memorial Because He Was Not Allowed To Speak, Lead Stories (2026)
- 6.Fact Check: Trump's Truth Social Account Did NOT Post Image Of Him Walking On Water With Caption 'Pastor. President. Protector.', Lead Stories (2026)
- 7.Did Trump's mom once call him an 'idiot with zero common sense'?, Snopes (2026)
Embed this verdict on your site
Paste this snippet to show our sourced verdict as a small card, with a link back to the full case file. Free to use.
<iframe src="https://theconspiratory.com/embed/counterfeit-social-posts" title="The Conspiratory verdict" width="520" height="190" style="border:0;max-width:100%" loading="lazy"></iframe>Help us investigate
This is a living case file. If you spot an error or know evidence we missed, tell us, and weigh in on where you land.
Where do you land?
Cast your read on this one.
Comments
Add your take. Comments are read and approved by a human before they appear, so keep it on topic and civil. Please do not accuse named, living people of crimes.