AI-generated image of Pentagon explosion causes market drop

[highplainsdem]

Source: The Independent

A fake image of an explosion near the Pentagon in Washington DC, which may have been created using AI technology, caused a brief market selloff.

The image purported to show a large cloud of smoke near the headquarters of the Department of Defense on Monday, and was shared by a string of online accounts on Twitter.

-snip-

The image, which Insider says has some characteristics of AI-generated images, was shared by the Twitter account of Russian-state media RT, among others.

The fake explosion was shared by the @Deltaone Twitter account at 10.06am ET, and by 10.10am ET the US stock market had fallen by 0/26 per cent before bouncing back, Insider reported. The account later posted that the explosion picture was fake.

-snip-

Read more: https://www.independent.co.uk/news/world/americas/fake-pentagon-explosion-market-drop-b2343709.html

---

Much more on this in the General Discussion thread at https://www.democraticunderground.com/100217935553 .

[Pacifist Patriot]

1. Oh fu....

[IthinkThereforeIAM]

2. That was fast...

... AI creating global havoc already.

[Comfortably_Numb]

3. JFC. I'm never going to be able to retire with these dipshits fucking the markets raw.

Potential World wide financial catastrophe because we can’t control fake shit on the web. Give me a god damn break.

[Wingus Dingus]

4. How about these market numbnuts wait for ACTUAL NEWS reports

before tanking the economy?

[C Moon]

12. Because that's not what they were told to do.

[paleotn]

5. Buying opportunity.

I wonder if the SEC is monitoring who piles into the market just after the tank?

[SouthernDem4ever]

7. They have algorithms that will pick up stuff out of the ordinary

don't know how bad it has to be for them to take action. Probably need proof.

[Shermann]

8. Better to target a single company's stock instead of the S&P

The dip wasn't anything to get rich quick on.

[durablend]

6. How long until someone "blows up" some oil infrastructure

Sending gas prices skyrocketing?

[LudwigPastorius]

9. ...and so it begins.

How does human civilization survive with no consensual reality?

[COL Mustard]

10. That's not the Pentagon.

This is the Pentagon:



The River Entrance is on the side closest to the camera. The Mall Entrance is to the right, where all the cars are.

[IronLionZion]

11. People should at least know that the Pentagon is pentagon shaped

pranksters on social media are a nuisance but people shouldn't fall for this nonsense so easily

[Initech]

13. Why am I not at all surprised that Russia is involved?

[Prairie_Seagull]

14. Our collective cynicism needs a significant adjustment

to the upside. Hell I thought mine was already at say 7 or so. What the hell will AI to do trust in the near future. How long before we question what we are reading on formerly trustworthy sites...

[honest.abe]

15. The market drop was cause by fake Twitter accounts reporting this as a real event.

Fake images like this were possible long before AI. Easily done with Photoshop.

[highplainsdem]

16. AI has made it much easier.

We're going to see a deluge of political deepfakes, thanks to AI. Even the people peddling and hyping AI are worried about that.

[Hugh_Lebowski]

17. Headline is still tenuous at best (nt)

[highplainsdem]

19. You can call it tenuous if you want. Most of the news stories

on this focused on AI as much as Twitter, for good reason. Did you see my GD post with the long Twitter thread from John Scott-Railton, who's a senior researcher at the Citizen Lab - https://en.m.wikipedia.org/wiki/Citizen_Lab - and researches spyware, phishing and disinformation? I linked to it in the OP.

Article from The Street there, too.

The fact that hoaxes were perpetrated before AI does not make AI's role in making them much worse now irrelevant or negligible. Even people who are real experts on AI don't downplay the risks the way some proponents of AI do online.

[Hugh_Lebowski]

20. I'm calling it that because it is exactly that given the actual case at hand

Article doesn't even say it's a definite AI-generated image.

And the spreading on Twitter/Social media is the actual cause.

[highplainsdem]

21. Again, see the GD thread. There were multiple tells that the hoax was

AI-generated, and Insider wasn't the only news source saying that.

[Hugh_Lebowski]

22. So AI created the image and created the story entirely of it's own accord?

And then created bots to spread the story all over the the internet?

Because that's what "the hoax was AI-generated" implies.

The CAUSE ... was actually a human, and a buncha bots. And credulous people believing it.

The image could've been generated without AI, as could have the text. Very easily. The AI aspect is basically immaterial, but being presented front and center in the headline.

That's why I say it's tenuous.

[highplainsdem]

23. I'm sorry you don't understand the role of AI here. But I'm glad journalists

and AI experts and lawmakers do.

Btw, AI-generated is the commonly used term for images and text - and hoaxes - produced by AI. And it has been for years.

https://openai.com/research/dall-e

https://openai.com/blog/new-ai-classifier-for-indicating-ai-written-text

https://www.insider.com/ai-generated-hoax-explosion-pentagon-viral-markets-dipped-2023-5

https://www.vice.com/en/article/7kx84b/ai-generated-pentagon-explosion-hoax-twitter


I don't know why you're trying to insist there's anything unusual or misleading about standard scientific and journalistic usage.

[Hugh_Lebowski]

24. Because I'm looking at the literal words of the headline and concluding it's tenuous

And you're the one who said the 'hoax' was AI generated, which I also replied to in my last post.

I can use AI to generate an image and story here on my PC, and it's not going to cause a market crash. It's not going to be a 'hoax'. The hoax is created by the person who leveraged the AI, and then used bots to spew it all over the internet. The person and the bots are the 'causes', IOW.

I don't know what's difficult to understand about this

[highplainsdem]

25. Look at the headlines of the stories at the last two links I just gave you.

Go quarrel with Insider and Vice and all the journalists and scientists who use the same wording if you want to. I'll continue with the standard usage, however much you want to complain about it.

[Hugh_Lebowski]

26. If you copied the headline exactly, then that's exactly who I'm quarreling with my friend :) (nt)

[Hugh_Lebowski]

27. BTW, Vice's headline is FAR more accurate

"Verified Twitter Accounts Spread AI-Generated Hoax of Pentagon Explosion"

I wouldn't complain for one second about that one. That one's (pretty) accurate

[honest.abe]

18. Yes agreed but to blame AI entirely is misleading.