To our readers: The importance of vetting photos in the age of AI
"If it sounds too good to be true, it probably is" needs a makeover.
The phrase for today should be, "If it LOOKS too good to be true, it probably is."
The emergence of ChatGPT and other generative artificial intelligence tools this year and the number of people playing with them to create images that don't exist in real life is a challenging development for news people.
Remember the quaint days when you could spot a bad Photoshop job by the telltale ragged edges of a cutout? I sure do.
Today, images created by artificial intelligence can be nearly indistinguishable from the real thing.
I talked about the challenges this poses with Jeff Knox, the Daily Herald's senior director of visuals.
"As a news organization we have to make sure what we publish is real. If it's not, we lose our credibility. We have a strict code of ethics that ensures what we publish is a real image and not altered to change it.
"Sometimes when we see an image that looks too good to be true or we are unsure of its source, we will contact the person who sent us the photo or run it through a Google image search to try to verify its authenticity."
Another way to help determine whether a photo was taken with a camera - and the extent to which it's been altered or enhanced - is to look at the metadata attached to its digital file. It's a finger print of sorts that tells you when a photo was taken and with what settings.
A recent investigative piece in The Washington Post concluded that there are AI-generated images on stock photo services, including Adobe Stock.
Users of such services could unwittingly be spreading fake images across the internet, believing them to be real.
The Post's reporting got Adobe Stock to say it will crack down on submissions of AI images as well as rid its catalog of existing ones.
It's bad enough that people use AI to create images and pass them off as real to stir your emotions and further an agenda, often a political one. Some fake images of the Israel-Hamas war, for instance, are indistinguishable from the real thing. And they're incredibly powerful.
So it behooves producers of news content, like us, to be doubly careful about the images we choose to use.
I remember many years ago we were including Associated Press wire coverage of a military bombing involving Russia. The AP carried a muddy black-and-white image of the strike, purportedly taken from the airplane. It was the only photo available.
My trust of the Russian government is fairly low, so I asked where it came from. The answer: a Russian government source. Better safe than sorry, I always say. So I opted not to run any photo at all.
Another side effect of the democratization of photo fakery is this: we will come to distrust the authenticity of every photo we see. As we do with social media already. As we do every time we see someone carrying a Louis Vuitton purse at the mall.
That is rapidly turning everyone into a cynic.
Just know that newspaper people are committed to portraying reality. Because, as Knox notes, we're nothing without our credibility.