News

Meta AI prompts you to choose to post publicly in the app's Discovery feed by defaultMeta has a new warning pop-up, but accidental sharing remains a possibilityYou can opt out of having your conversations go public entirely through the Meta AI app’s settingsThe Meta AI app's somewhat unique contribution to the AI chatbot app space is the Discovery feed, which allows people to show off the interesting things they are doing with the AI assistant.However, it turns out that many people were unaware that they weren't just posting those prompts and conversation snippets for themselves or their friends to see. When you tap "Share" and "Post to feed," you're sharing those chats with everyone, much like a public Facebook post.The Discovery feed is an oddity in some ways, a graft of the AI chatbot experience on a more classic social media structure. You’ll find AI-generated images of surprisingly human robots, terribly designed inspirational quote images, and more than a few examples of the kind of prompts the average person does not want just anyone seeing.I've scrolled past people asking Meta AI to explain their anxiety dreams, draft eulogies, and brainstorm wedding proposals. It's voyeuristic, and not in the performative way of most social media; it's real and personal.It seems that many people assumed sharing those posts was more like saving them for later perusal, rather than offering the world a peek at whatever awkward experiments with the AI you are conducting. Meta has hastily added a new pop-up warning to the process, making it clear that anything you post is public, visible to everyone, and may even appear elsewhere on Meta platforms.If that warning doesn't seem enough to ensure your AI privacy on the app, you can opt out of the Discovery feed completely. Here's how to ensure your chats aren’t one accidental tap away from public display.(Image credit: Meta AI screenshot)Open the Meta AI app.Tap your profile picture or initials, whichever represents your digital self.Tap on "Data and Privacy" and "Manage Your Information."Tao on "Make all public prompts visible to only you," and then "Apply to all" in the pop-up. This will ensure that when you share a prompt, only you will be able to see it.If that doesn't seem like enough, you can completely erase the record of any interaction you've had with Meta AI by tapping "Delete all prompts." That includes any prompt you've written, regardless of whether it's been posted, so be certain.Private promptsOf course, even with the opt-out enabled and your conversations with Meta AI no longer public, Meta still retains the right to use your chats to improve its models.It's common among all the big AI providers. That's supposedly anonymized and doesn't involve essentially publishing your private messages, but theoretically, what you and Meta AI say to each other could appear in a chat with someone else entirely in some form.It's a paradox in that the more data AI models have, the better they perform, but people are reluctant to share too much with an algorithm. There was a minor furor when, for a brief period, ChatGPT conversations became visible to other users under certain conditions. It's the other edge of the ubiquitous “we may use your data to improve our systems” statement in every terms of service.Meta’s Discovery feed simply removes the mask, inviting you to post and making it easy for others to see. AI systems are evolving faster than our understanding of them, hence the constant drumbeat about transparency. The idea is that the average user, unaware of the hidden complexities of AI, should be informed of how their data is being saved and used.However, given how most companies typically address these kinds of issues, Meta is likely to stick to its strategy of fine-tuning its privacy options in response to user outcry. And maybe remember that if you’re going to tell your deepest dreams to an AI chatbot, make sure it’s not going to share the details with the world.You might also likeI tried the new Meta AI app and it's like ChatGPT for people who like to overshareMark Zuckerberg wants everyone to have AI friends, but I think he's missing the point of AI, and the point of friendshipMeta AI is now the friend that remembers everything about youMeta wants to fill your social media feeds with bots – here's why I think it's wrong

Google’s Veo 3 Fast generates 720p AI videos twice as fast as standardGemini Pro users can create three videos per dayFlow Pro users pay just 20 credits per clipThere's a strawberry made of glass that someone is cutting like it's made of jelly, then the same thing happens to several other berries and the Pokémon Charizard. A woman dips tongs into a platter of molten rock and takes a bite of the apparently delicious treat before spreading some on a waffle and taking a bite.These aren't a cough-syrup-induced hallucination; it's the latest trend in ASMR videos, created with Google's Veo 3 and other AI movie generators.You might have seen these and other bizarre videos on your TikTok algorithm. I’m not the biggest proponent of ASMR videos and their gentle whisperers, rhythmic tapping, and other soporific audio, but I understood the appeal. I'm not so sure the AI version is just as good. These aren’t your traditional low-fi lo-fi tapping-and-brushing videos. @softcrunchai ♬ original sound - SoftCrunchAI Google Veo is definitely the most popular choice of AI video generators for ASMR. As good as the model is at producing realistic videos (for a given value of realism in this case), it still creates videos with a sheen of artificiality, lacking the errors and imprecision that are the hallmark of human-made ASMR.Friends of mine who are much bigger fans of ASMR claim it's not just the sounds and voices that entice them. It's the intimacy and immediacy that they like. One said that the 'tingle triggers' are there in the AI videos, but it's just not the same.AI ASMR @impossibleais ♬ green to blue (slowed + reverbed) - daniel.mp3 That's not a universal opinion, though. These videos rack up millions of views. The comments are full of both excitement and confusion, with people unable to explain why they like the videos, or who are amazed how they can't stop watching them.Part of the appeal might be novelty. Anything new will have a fanbase at least for a little while, and Veo’s video quality is unlike anything most people have seen before. It mimics natural lighting, shadow, and realistic camera motion. Perfect glass statues of Pokémon that can be cut in half with ease, because they're all AI-generated, are an eye-catcher for sure.As fun as it is, I wonder if anyone seeking a pure ASMR experience will choose an AI video. The sounds may scratch that auditory itch, but I wonder how many lists of favorite ASMR videos will include them.Not every AI trend has to make sense. Some of them very much don't. AI ASMR may have a niche place in the overall ASMR ecosystem. Still, I don't think the majority of people who fall asleep to sweet whispers will prefer the sound of simulated glass crunching under an invisible knife. @impossibleais ♬ September - slowed + reverb - Wallerstedt You might also likeTikTok owner ByteDance has a new AI video creator you have to see to believe5 of the most heartwarming uses of AI we’ve seen so farSora needs to up its game to match the new Runway AI video model