AI

TechCrunch Minute: Meta is labeling real photos as ‘Made with AI’

Comment

Photographers are claiming that Meta is tagging their real photos as “Made with AI.” 

Meta introduced this automated tag to help people understand when what they’re seeing is real and when it’s been computer generated. But even AI itself isn’t good at telling what is or is not AI. So, users like a former White House photographer and a professional cricket team end up posting photos that are flagged as being AI-generated.

Former White House photographer Pete Souza told TechCrunch that he thinks this might be a result of basic photo editing that photographers do before uploading a photo. So, if you upload an image into Photoshop, crop it, and then export it as a new .JPG file, Instagram’s detectors might go off. This makes sense, since Instagram says its AI looks at metadata to detect the legitimacy of an image.

As election season approaches, it will become increasingly important that massive social platforms like Meta can adequately moderate AI-generated content. But as it stands, does a “Made with AI” tag really accomplish anything if it’s not reliable?

More TechCrunch

Is AI going to make us all more productive at work? Executives are feeling optimistic, but workers aren’t quite as convinced. That’s according to a…

TechCrunch Minute: Is AI actually making workers more productive?

Live sports are finally moving onto streaming in a serious way, with a new service called Venu Sports set to launch this fall. This is…

TechCrunch Minute: With Venu Sports, streaming gets serious about live sports

For once, the iPad was the highlight of Apple’s latest sales numbers. The company said in its third-quarter earnings report that iPad sales increased year-over-year,…

TechCrunch Minute: The iPad was the surprising star of Apple’s sales numbers

As widely reported, Wiz recently said no to a $23 billion acquisition offer from Google. At that price, it would have been Google’s biggest acquisition…

TechCrunch Minute: Why did Wiz walk away from $23 billion?

Startups are attracting serious funding to provide fun, alcohol-free drink options at the bar and in other social contexts. Earlier this year, canned water company…

TechCrunch Minute: Investors pour money into non-alcoholic beverages

Although AI models have made big breakthroughs in the last few years, there’s now a risk that they could get dumber and even collapse, according…

TechCrunch Minute: Will AI get dumber?

It turns out that maybe, just maybe, spending a lot of time staring at your phone and reading bad news isn’t great for your outlook…

TechCrunch Minute: Doomscrolling probably isn’t great for your mental health

Bing is the latest search engine to introduce generative AI into its results, with Microsoft announcing that it’s making Bing generative search available to a…

TechCrunch Minute: Microsoft brings more generative AI to Bing

Meta’s AI integrations into apps like Instagram and Facebook have felt pretty useless so far. If I’m just trying to find my friend’s Instagram, why…

TechCrunch Minute: Meta’s new ‘Imagine me’ feature lets you generate AI selfies

Today’s story takes us to a high school on the outskirts of Essex, which has ended up on the wrong side of the U.K.’s GDPR…

TechCrunch Minute: A UK school was reprimanded for unlawful use of facial-recognition technology

This weekend, President Biden announced that he would suspend his campaign for president, and he endorsed his vice president, Kamala Harris. We still don’t know…

TechCrunch Minute: Where Kamala Harris stands on tech issues

Late last week, there was a worldwide tech outage that affected everything from airports to banks to healthcare. Flights were grounded across the United States.…

TechCrunch Minute: What caused last week’s major tech outage?

OpenAI is launching a mini version of its latest AI model. These small AI models are meant to be faster and more affordable than the…

TechCrunch Minute: OpenAI shrinks its flagship model

What do MrBeast, John Oliver and The Wall Street Journal have in common? The transcripts of their YouTube videos have been scraped to train the…

TechCrunch Minute: Over 100K YouTube videos have been scraped to train AI for Apple, Nvidia

If you play Wordle every day but also happen to enjoy chess, we’ve got great news for you.  Echo Chess, a game built by a…

TechCrunch Minute: This startup wants to make Wordle for chess

Trouble could be on the horizon for OpenAI. A group of whistleblowers accused OpenAI of requiring its employees to sign illegally restrictive non-disclosure agreements, or…

TechCrunch Minute: Whistleblowers say OpenAI employs ‘illegally restrictive’ NDAs

Google found a new way to demonstrate what its Gemini AI model can do, with help from a robot. This was a robot from Google’s…

TechCrunch Minute: A Google robot shows off what Gemini can do

A new study from the Federal Trade Commission looks at the “dark patterns” that subscription apps and websites use to get more money or data…

TechCrunch Minute: FTC finds ‘dark patterns’ in subscription apps

It’s a big day in the Samsung universe. At Samsung’s Unpacked 2024 event, the company debuted new phones, headphones and even a wearable ring.  As…

TechCrunch Minute: Unpacking Samsung Unpacked

Not gonna lie, things aren’t looking great for NGL.  The Federal Trade Commission (FTC) is banning NGL from serving minors or advertising to them, marking…

TechCrunch Minute: The FTC bans anonymous social app NGL from serving minors

Florida man is back with a vengeance, and this time, he’s going after Walmart delivery drones.  Yes, really. Walmart was demonstrating its delivery drone technology…

TechCrunch Minute: This is what happens when you shoot down a delivery drone

You probably like to think of yourself as reasonably sophisticated when it comes to online scams — you’re not opening random email attachments or sending…

TechCrunch Minute: How to protect yourself from AI scams

It’s easier than ever to make a deepfake of another person’s voice or face, but at least YouTube is making some small changes that make…

TechCrunch Minute: YouTube makes it easier to report and take down AI deepfakes

“Fediverse” is one of those terms that gets thrown around a lot, but what does it actually mean? The word comes from “federation” and “universe”…

TechCrunch Minute: A beginner’s guide to the fediverse

Will robots ever have living, human skin? Researchers have been exploring the question and they’ve come up with some pretty unsettling results. I think it’s…

TechCrunch Minute: Beware the smiling robot with living skin

Reddit is taking a stand against AI companies — or at least asking them to pay up. Earlier this week, Reddit announced that it’s changing…

TechCrunch Minute: Reddit is taking a stand against AI crawlers

X still has a verified bot problem. Also, grass is green.  We knew it would be a bad idea to let anyone pay for a…

TechCrunch Minute: X still can’t handle its bot problem

A former Snap engineer created an uncanny valley for social media, and it’s an app called Butterflies.  Butterflies looks like Instagram, except it’s a platform…

TechCrunch Minute: Former Snap engineer made a social network for AI

OpenAI rival Anthropic released its newest generative AI model this week, Claude 3.5 Sonnet. While Sonnet is an improvement over Anthropic’s last release, it’s not…

TechCrunch Minute: Anthropic’s Claude 3.5 is its best model yet — at least on paper

Hey, tech detox enthusiasts! If you’re tired of your phone being smarter than you, meet the Light Phone III. It’s a minimalist marvel with an…

TechCrunch Minute: The Light Phone III is light on features but heavy on price-tag