
Hoax fighter and fact finder. Main tools: reverse image search and coffee. Operated by @JanneFI Bluesky: https://teapawt.net/t.co/6hFrpe68eR
Joined April 2015
- Tweets68.6K
- Following489
- Followers185K
- Likes65.5K
HoaxEye retweeted
Creating fake, AI-generated images of Auschwitz and other camps is a dangerous act of distortion. Presenting them as authentic photographs falsifies history and disrespects the memory of the victims. Social platforms, especially Facebook, bear responsibility for enabling and amplifying their spread.
These examples come from the Facebook page “Lost in Auschwitz.” A “then and now” comparison promises a connection between historical documentation and the authentic site today. Here, both sides are AI-generated fakes. Placed side by side, the two fabrications appear to validate each other: the supposed present makes the invented past look credible, while the supposed past gives the invented present historical meaning.
Photography carries an expectation that someone stood in a real place and recorded what was there. This trust is fundamental to the use of photographs as historical sources. Authentic photographs of Auschwitz are evidence. Photographs taken at the Memorial today also document a real place whose surviving buildings and ruins bear witness to the crime.
AI-generated images exploit this trust. They imitate the visual language of documentary photography without providing its connection to reality. Labels such as “THEN 1944” and “NOW 2025” reinforce the false impression that viewers are looking at photographic records.
Authentic photographs, documents, testimonies, objects and the preserved site allow us to study the history of Auschwitz and remember its victims. Replacing them with plausible-looking inventions erodes the distinction between evidence and fiction.
We believe that platforms have a responsibility to intervene when they are used to spread such falsified content. Allowing AI-generated fakes of Auschwitz to circulate degrades the dignity of the victims, endangers public understanding of history, and, in fact, supports the distortion of history.
The history of Auschwitz and the Memorial that preserves all its remains must not be replaced with AI-generated fiction. Memory and historical truth deserve much stronger protection.
You might have seen a video of Palestinian and Jordanian flags being flown from the top of the Eiffel Tower recently.
But, this clip was generated with AI and there’s no evidence it shows a real event.
Read our check here 👇
fullfact.org/world/palestine…
HoaxEye retweeted
‘Russia is not an enemy’, says a Russian official channel.
Meanwhile, European authorities investigate arson, sabotage and covert operations attributed to Russian intelligence.
For 10 weeks, we asked five AI chatbots the same questions every day.
None repeated the core misinformation they were tested on, but they reported proposals as law, quoted outdated figures and gave different answers on different days.
Our first report: fullfact.org/ai-benchmark/re…

ALT Millions of people increasingly rely on AI chatbots for timely responses to factual questions. Our findings show the answer comes back in the same confident tone whether it is correct, out of date, or describes something that has not yet happened. That is the real problem here. These models are not necessarily peddling misinformation, but their responses can be incomplete, lack context or be poorly sourced. The bosses of these AI companies say they are happy for others to mark their homework, so we have started doing it for them.
HoaxEye retweeted
Wildfires in Indonesia have shrouded parts of the region in toxic haze, but satellite imagery and Malaysian authorities indicate a circulating map exaggerates the scale of the fires. We also found it was generated with AI. u.afp.com/SCQW
HoaxEye retweeted
🌀 No human-made technology is capable of creating or redirecting hurricanes.
Despite their near-simultaneous development, meteorologists say tropical systems Isaias, Rachel and Simon resulted from natural weather conditions u.afp.com/SChy
Now that the VASCO 2025 sample (107K rows) is finally public, I could comment the Machine Learning paper and some other topics more broadly. Please see my Metabunk post: metabunk.org/threads/transie…
Some key points:
- the published ML results can't currently be regenerated from the published artifacts
- the 107K catalogue can't be derived from the method the 2022 paper describes
- I tested POSS-II red plates, same sky, at an epoch 35–45 years later, with human satellites in orbit. Results are interesting
Double trouble -- our peer-reviewed machine learning/transient paper has also just been published. This is where we clean up the sample and see that the fewer contaminants there are, the stronger the correlation with nukes and the larger the shadow deficit. It also shows that the correlations with nukes is strongest 24 hours before a nuclear test.
Link: nature.com/articles/s41598-0…
If you have time to read only one document, read this: "Four things the published release lets anyone check" github.com/jannefi/poss1-pla… I'm pointing out few findings anyone can check. No finger-pointing is involved - or needed
4
2,061
HoaxEye retweeted
Fact Check: FAKE Secretly Recorded Footage Shows Elon Musk Observing A Portal At SpaceX -- AI
youtube.com/watch?v=frKGhfmQ…
A video on Facebook claims that new lockdowns might be needed to deal with a dramatic rise in Covid-19 cases in the past week.
But official data shows only a small rise in cases, which remain at a very low level.
fullfact.org/health/covid-lo…
We’ve been trialling a tool comparing the output of various LLMs over time to questions about false claims circulating online. Here are the results 👇 fullfact.org/technology/full…
HoaxEye retweeted
As rescuers in Nepal and China searched for thousands missing after deadly flash floods swept through the Himalayas, social media was inundated with clips falsely linked to the disaster.
Here are some of the claims our global team has debunked so far 🧵
HoaxEye retweeted
This video, viewed nearly 6 million times, is clearly AI-generated.
Apart from multiple errors and inconsistencies visible in the video, Google's SynthID tool indicates that it was created with Google AI.
😭😭😭 कसैले कल्पना नै गरेको थिएन यस्तो हुन्छ भनेर… प्रकृति बौलाएपछि आखिर कसको के नै लाग्दो रहेछ र! 🌊💔
यस्तो विपत्तिमा परेका सबैको सकुशल उद्धार होस्। 🙏😭
Readers added context they thought people might want to know
This is AI generated. The panning of the purported mobile phone is unnatural as well as the forces and movements. Buildings and trees suddenly appear in the second panning revealing the fake video.
aapnews.aap.com.au/news/fake-floo…
HoaxEye retweeted
There's no evidence that this is what they were used for, they also don't work, it's been tried.
It's also only found in only one street.
As a archaeologist currently working in Pompeii said:
"It’s such complete nonsense "
"Anything you also see in this video is fictional and absolutely never happened" Source: youtube.com/watch?v=q_dzmB5f…
HoaxEye retweeted
This video, purporting to show US Congresswoman Nancy Mace dancing erratically in a bar and contorting her face at the camera, is a doctored fake.
It is altered from a clip of a comedian.
factcheck.afp.com/doc.afp.co…
That does not seem to be breaking news. The photo is from 2020: bbc.com/pidgin/world-5367096…




























