Philip Eil is a freelance journalist based in Providence. His debut book is “Prescription for Pain: How a Once-Promising Doctor Became the ‘Pill Mill Killer.’”
In the early days of AI-generated images and videos, there were telltale signs that the content wasn’t real.
Extra limbs. Garbled text. Missing fingers. A gauzy, airbrushed-looking background.
But recently, with little fanfare, we entered a new era. The giveaways are rarer and subtler, if they exist at all.
There is a growing consensus among AI analysts: The era of regular people being able to spot AI fakes unassisted is over.
To put that another way: As you navigate the digital world — on websites, social media, group chats, and elsewhere — you have good reason to question whether what you’re looking at is real.
We don’t need to speculate about what happens in a world inundated with convincing AI fakes. It’s already here.
Consider a few events from the first half of 2026.
In February, the release of a hyperrealistic 15-second AI-generated clip of a cinematic fistfight between Brad Pitt and Tom Cruise sent shock waves through Hollywood and elicited a stern statement from the Motion Picture Association rebuking ByteDance, the China-based owner of TikTok and SeeDance 2.0, the AI video generator used to create the clip.
In March, Israel’s Prime Minister Benjamin Netanyahu began a press conference by stating, “First of all, I just want to say I’m alive, and you’re all witnesses,” after rumors circulated that he had died and social media users continued to debate whether videos of him online looking healthy were AI-generated. (There’s no evidence that they were, but many skeptics remain unconvinced).
At other points this year, we have seen news outlets duped by AI-generated images. US elected officials “liked” fake images purporting to depict a rescued American pilot in Iran and shared fake images of the capture of Venezuelan head of state Nicholas Maduro. The White House shared an AI-doctored video of a Team USA athlete saying disparaging things about Canada that he never actually said.
These kinds of events will almost certainly become more common, because of the extraordinary speed with which AI technology is advancing. At the same time, tools to detect fake images are underwhelming.
No major federal laws have been passed to regulate the dissemination of AI-generated imagery. And given the Trump administration’s bullishness on AI, that’s unlikely to change. State legislatures have been nibbling around the edges of the issue — regulating deepfake pornography and AI-generated child sexual abuse material or the use of AI-generated images in political campaigns — but it’s unclear how aggressive they can or will be. In Massachusetts, where AI fakes are increasingly visible in campaign materials, two bills to regulate them passed in the House and await votes in the Senate.
Meanwhile, tech giants are gesturing toward addressing the issue via a consortium called the Coalition for Content Provenance and Authenticity, which aims to someday establish a “nutrition label for digital content.” But in practice, most platforms remain unregulated free-for-alls.
A telling moment came in February, when the Auschwitz Memorial blasted Meta for its inaction on AI-generated images of the Holocaust that have proliferated on Facebook. “By allowing these distortions to surface, circulate, and gain visibility, Facebook directly contributes to the erosion of factual understanding of the complex history of Auschwitz, which we try to protect,” it said.
That moment underscored a grim reality: When it comes to AI fakes, nobody is coming to save us. For now, it’s on regular folks to determine what’s real and what isn’t.
So what should you do?
My first bit of advice is simple: If you’re online and you’re anywhere other than a credible news site, you should approach content with caution. And while you may occasionally get lucky, you should not trust your own abilities to spot an AI-generated fake.
This is not a marker of your gullibility or intelligence; it’s a reflection of just how sophisticated this technology has gotten.
I would also advise equipping yourself for today’s online info-chaos. Add a bookmark to the News Literacy Project’s “RumorGuard” page and the “Digital Forensics” page from Toronto Metropolitan University’s “Deepfakes Tracker.” Get acquainted with the fact-checking pages of news organizations like AFP and Reuters, which have made a habit of debunking AI-generated content. Check out books like “Verified: How to Think Straight, Get Duped Less, and Make Better Decisions About What to Believe Online,” from two leading digital literacy scholars, Mike Caulfield and Sam Wineburg. Ask your local librarian for tips on staying sane in today’s media landscape. The AI fakes may be more abundant and convincing than ever, but media literacy still works. And in today’s information environment, it’s a cornerstone of responsible citizenship.
A third, related tip, would be to not give up on the truth. I know that it’s harder, slower, and more expensive to find out what’s really going on. I know we’re all tired — so tired. But when folks give up on believing anything, it spells doom for democracy. I had a lot of conversations for this article, and something that I heard with striking regularity was fears about what happens when people start doubting everything and believing nothing. “That is the thing that keeps me up at night,” Renee Hobbs, the founder of the Media Education Lab, told me.
Some experts advise people to simply spend less time on social media. In a 2025 TED talk, Hany Farid, a Dartmouth professor who has been called “the godfather of digital forensics,” described social media as a place created by Silicon Valley to steal your time and your attention by delivering you the equivalent of junk food. Like any bad habit, you should quit social media, he said. And if you can’t quit, at least don’t use social media as a source of information. “Because it is simply too riddled with lies and conspiracies and now AI slop to be even close to being reliable,” he said.
But if you can’t quit, then there are ways to be more intentional about what content you consume. Jeremy Carrasco, who runs the popular Instagram account @jeremyfindsAI, draws a distinction between accounts you intentionally follow and the algorithm-driven and increasingly AI-drenched “For You” pages of platforms like YouTube, TikTok, and Instagram. He recently advised his followers to err on the side of assuming that nothing served to them by a “For You” page is real and urged them not to let those pages inform their views on reality.
One person I spoke to had a memorable bit of practical, albeit chilling, advice. Manny Ahmed is the cofounder of the UK-based digital media verification company OpenOrigins. I contacted him after seeing an April 2024 interview in which he predicted that in about two years — in other words, right now — photos would be as easily fabricated as text.
During our chat, we briefly touched on advances in AI-generated audio content, which prompted Ahmed to remind me that if you get a call from a loved one claiming to have been kidnapped and asking for a ransom, you shouldn’t believe it.
To prepare for such scams, he urges people to establish a password or code phrase with family members that they can use to verify that the voice that sounds like them actually is them.
Be sure to make that password or catchphrase something that an AI bot can’t fake, he says.
“Because AI can definitely fake your voice.”