Picture this. You snap a group photo at a concert but the edge cuts off half the stage.

Instead of tossing the shot, your phone could soon add new scenery on any side to rescue it.

That is the new extend feature coming to iPhones with iOS twenty seven.

Think of it like autocomplete on your phone keyboard, except the guesses fill in missing parts of a picture rather than words.

The system studies the colors, shapes, and patterns already in your photo, then continues those patterns outward in a way that usually feels natural.

It works on the top, bottom, left, or right, so you can fix a cropped horizon or make a tight selfie feel more open.

Almost every teen takes and shares photos daily, so this removes a tiny but annoying limit on creative sharing or school projects.

When the update lands you simply open the Photos app, pick an image, drag the new extend handles, and watch the blank space fill in right away.

No extra apps or accounts are required.

The real magic happens because the AI treats the photo like a puzzle where it already knows most of the pieces.

It predicts what the missing pieces should look like based on everything it has learned from millions of other images.

That same idea of guessing what comes next powers many tools we use every day, from texting to music recommendations.

And that is basically how the extend feature works, not so scary right.

One way to picture the prediction step is to imagine a painter who only sees the left half of a landscape but still finishes the right half so it matches perfectly.

Another angle is a jigsaw puzzle where the AI does not just copy pieces, it invents new ones that keep the same lighting and style.

A third view is like a band that hears the first verse of a song and then keeps playing in the exact same rhythm and mood.

These layers show why the results often look natural instead of obviously added.

Apple's Image Playground also received major updates this week. The refreshed tool upgrades from rigid voice commands to a system that can chat back and forth while remembering earlier context. It combines smaller on-device models with larger cloud models for added power when needed.

Most users already rely on Siri for timers and messages, so the conversational shift could reshape daily phone interactions during homework or planning. The features are slated for fall release on compatible iPhones. Ars Technica noted the change moves AI from a separate app into a built-in companion.

This development continues Apple's ongoing push to embed smarter assistance directly in hardware people carry every day.

OpenAI expanded free-tier image generation in ChatGPT, giving non-paying users more creation power without upgrades. K-pop fans responded by highlighting realistic deepfake videos and images made without consent. These fakes often place idols in awkward or misleading scenes that spread quickly on social platforms. Reddit threads documented the calls for better detection tools and awareness.

The episode stressed that spotting AI media is turning into a routine skill for everyday scrolling. One practical step is comparing suspect images side-by-side with verified photos on trusted sites. This thread shows how rapid image tools create both creative access and new moderation challenges.

Okay, now for my favorite part of the show where we turn that understanding into action you can take right now.

First up is a free tool from Deezer that scans playlists on Spotify or Apple Music to flag tracks made by AI.

More music online is created by computers, and many listeners want to know what is human made versus machine made.

Go to the Deezer website, find the AI detector tool, paste a playlist link from another service, and see what it flags.

Next, DoorDash added an AI chatbot that lets you order food by typing a description or even snapping a photo of a menu.

The chatbot reads the photo or your words and turns them into an order without you scrolling through endless options.

Try it by opening the DoorDash app, tapping the chat option, and either typing what you want or uploading a picture of a dish you saw online.

Finally, Google released live speech to speech translation that works almost instantly during video calls.

Think of it like having a super fast interpreter in your pocket who listens, understands, and speaks the translation while the first person is still talking.

It keeps the original speaker's tone and rhythm so the voice still sounds like the same person.

Students could practice conversations with people around the world without waiting for subtitles.

To try the translation side, keep an eye on Google apps this summer and test it during a call with a friend who speaks another language.

Here are a few quick bits that caught my eye this week.

One AI researcher suggests saving your hardest or most unusual ideas now because AI is making it cheaper to turn good ideas into finished projects.

The bottleneck is shifting from how do I build this to what should I build.

Just a year ago the most advanced AI helpers could only answer questions.

Today they can take actions across multiple tools, which is exciting but also means we need to think carefully about safety as they get more powerful.

A separate report found that some image tools are still hosting problematic deepfake images even after earlier complaints, which shows why learning to spot AI generated media is becoming an everyday skill.

You can practice by searching for AI deepfake examples on a trusted site and comparing them side by side with real photos to train your eye.

Visa is testing connections between its payment system and ChatGPT so AI agents can eventually complete purchases with user permission.

The effort remains early yet signals how financial firms are preparing for agents that handle routine buying.

Mistral is reportedly moving away from frontier model development toward contractor-style work according to recent posts.

Related commentary notes that non-key tasks may gain the most from stronger models while assumptions about routing simple jobs to smaller ones may be flawed.

Google Earth flight simulation surfaced as a lighter alternative to complex fable-style games in community discussions.

That is it for today. Remember, every AI expert started exactly where you are right now.

If something we talked about today made you curious, go try it, that is literally how learning works.

Stay curious, keep experimenting, and we'll see you tomorrow.

And before you go, this show is part of the Nerra Network, a family of daily podcasts covering tech, science, markets, and more.

If you enjoyed today's episode, give Models and Agents a listen: your daily briefing on AI models and agents.

You can explore the whole lineup at nerranetwork.com.

That's it for today! Remember, every AI expert started exactly where you are right now. If something we talked about today made you curious, go try it — that's literally how learning works. Stay curious, keep experimenting, and we'll see you tomorrow. And before you go — this show is part of the Nerra Network, a family of daily podcasts covering tech, science, markets, and more.

If you enjoyed today's episode, give Models and Agents a listen: your daily briefing on AI models and agents. You can explore the whole lineup at nerranetwork.com.

Sources

• reddit.com
• engadget.com
• reddit.com
• x.com
• arstechnica.com
• x.com
• techcrunch.com
• reddit.com
• x.com
• reddit.com
• wired.com
• reddit.com
• the-decoder.com
• theverge.com
• arstechnica.com
• engadget.com
• technologyreview.com
• techcrunch.com
• x.com
• wired.com
• Google News
• x.com
• reddit.com
• theverge.com
• reddit.com
• x.com
• x.com
• x.com
• x.com
• x.com
• x.com

💬 Reply to this email — Patrick reads every one.

Share: X · LinkedIn · WhatsApp

Forwarded this email? Subscribe here — it's free.