Verified Twitter Account Uses AI to Spread Hoax Pentagon Explosion

There is always a dark side to any piece of tech. AI is no exception.

Dozens of verified Twitter accounts with large numbers of followers showed this picture today.

Only the photo posted on Twitter did not have a big red FAKE on it.

Accounts such as @WarMonitors, @BloombergFeed, and RT posted an image of a large, gray smoke cloud appearing next to a white government building with a corresponding caption that stated there was an explosion near the Pentagon.

Not surprisingly, when this showed up on Twitter, it got retweeted, including other “verified” accounts.

After an hour, the Arlington, Virginia fire department tweeted that there is no explosion or incident at or near the Pentagon. That took an hour.

Some of these accounts have up to a million followers.

And blue check marks.

This is similar to other “verified” Twitter accounts saying that Russian military jets were being armed with nuclear payloads. That happened earlier this month.

This creates two issues.

All a blue check mark means at this point that the user behind the account has paid to have a checkmark. Nothing more.

The other is that you cannot believe what you see anymore. AI generated images, including one from last month that claimed to show Trump in prison orange.

The stock market took a hit when this was posted but did recover after the fire department said this was fake.

This particular photo was trivial to verify.

What if the AI generated photo showed a candidate for high political office doing something illegal. Trump famously said that he could shoot someone on Fifth Avenue and get away with it. What if someone generated AI video of him doing that. Again, that one might be relatively easy to verify.

But what if the AI generated image was not easy to verify as a fake. Say, perhaps, pictures of some official in a compromising situation. How do you verify whether that compromising situation is real or not.

Social media companies are always going to be behind the power curve here.

And, AI generated images are only going to get more realistic every day, making it even harder to detect fakes.

At this point, all you can do is question what you see. Unfortunately, a lot of people are going to believe what they want to see.

This portends what next year’s presidential election will look like. Even if the candidates themselves don’t do it, people aligned with the candidates will do it. I predict.

Credit: Vice Media

by