Here's the part that should stop you mid-scroll: someone can type your Instagram username into a Meta AI prompt right now, and the system will generate an image using your face and your photos — no permission required, no notification sent to you, ever.
Meta quietly rolled this out alongside Muse Image, its new AI image generation model launched through what the company is calling Meta Superintelligence Labs. The model is baked directly into Instagram, and the integration is as seamless as it is unsettling. If your account is public — and most people's are — you're already opted in by default.
The feature is framed by Meta as a fun, creative personalization tool. Want to mock up a custom event invite with a friend's likeness? Generate a stylized graphic featuring a colleague? Just tag their handle. Meta AI pulls from their public posts and builds something new. The company clearly sees this as a feature. A lot of users are going to see it as a violation.
What makes this particularly thorny isn't just the opt-out model — annoying as that is. It's the silence. Meta's own help documentation confirms that users will not receive any notification when someone creates AI-generated content using their photos. You could have a hundred AI-generated images of your face floating around Instagram right now and have absolutely no idea.
The opt-out process exists, but it's buried. You'll need to open Instagram, navigate to your profile, tap the three-line menu, scroll to a section called Sharing and Reuse, and manually toggle off permissions for both Posts and Reels. That's four steps to protect something most people didn't know was at risk in the first place.
There's also no retroactive fix. If someone already generated an AI image using your content before you found the toggle, that image stays up. Turning off the setting prevents future generations, but the past is the past.
This fits a depressingly familiar pattern across the tech industry. Google recently updated its reverse image search to retain uploaded photos for AI training purposes — also opt-out, also not prominently advertised. The default is always participation. The burden is always on the user to go looking for the exit.
The timing is notable too. Meta is clearly racing to catch up with OpenAI and Google in the AI image generation space, and it has a massive built-in advantage: billions of photos uploaded by real people over nearly two decades. Muse Image isn't just a new product launch. It's Meta finally monetizing that archive in the most direct way possible.
If you're on Instagram with a public account, the five minutes it takes to find that settings toggle is worth your time. And if you've been thinking about switching to private anyway, consider this the nudge.
POLICY
AI Data Centers Are Crushing Trump's Made in America Energy Plans
A 141-year-old brick company in Ohio has watched its monthly electricity bill jump from $1,600 to $12,000 — not because it's running more kilns or hiring more workers, but because the power grid it depends on is being overwhelmed by AI data centers.
That's the increasingly hard-to-ignore collision happening across the Rust Belt right now. The same administration championing a manufacturing revival is simultaneously turbocharging the AI industry, and those two agendas are starting to eat each other alive on the electric grid.
The numbers from PJM Interconnection, the grid operator covering 13 states from Illinois to New Jersey, tell the story plainly. Capacity prices — what gets paid to power generators based on supply and demand projections — have gone from roughly $29 per megawatt-day in 2024 to over $329 per megawatt-day in 2026. That's not a gradual increase. That's a structural shift in who the grid is being built to serve.
For steel manufacturers, this is an existential squeeze. Electricity makes up somewhere between 20 and 40 percent of total steel production costs, and the electric arc furnaces used in modern steelmaking are enormous power consumers — each one drawing anywhere from 40 to 200 megawatts. Ohio-based steelmaker Metallus says its electricity costs have risen 70 percent since 2024, translating to an extra $15 million per year just to keep the lights on.
The irony cuts deep. US steelmakers have actually benefited from the data center construction boom, since building those massive facilities requires an estimated one million tons of steel annually. The industry is simultaneously getting a demand boost and an operating cost spike from the same source. The net math is not flattering.
PJM's own forecasts project that electricity demand in its territory will outpace available supply by 6.6 gigawatts starting in 2027. To put that in terms that land, that's the output equivalent of more than six nuclear power plants worth of missing capacity. Steel industry executives are already warning that production outages could follow if local grids buckle under the pressure.
Manufacturers are responding the only ways they can: raising prices, absorbing losses, or in some cases, looking at relocating operations entirely. None of those outcomes align with a domestic manufacturing revival narrative.
The political tension here is genuinely awkward. The Trump administration has leaned heavily on tariffs and trade pressure to push manufacturing back to American soil, while simultaneously positioning the US as the global leader in AI development. Both goals require enormous amounts of energy. The grid, at least in the industrial Midwest, is not big enough for both right now.
Something has to give — and based on current trajectory, it looks like the brick factories and steel mills will give before the data centers do.
Source: Ars Technica
That's the increasingly hard-to-ignore collision happening across the Rust Belt right now. The same administration championing a manufacturing revival is simultaneously turbocharging the AI industry, and those two agendas are starting to eat each other alive on the electric grid.
The numbers from PJM Interconnection, the grid operator covering 13 states from Illinois to New Jersey, tell the story plainly. Capacity prices — what gets paid to power generators based on supply and demand projections — have gone from roughly $29 per megawatt-day in 2024 to over $329 per megawatt-day in 2026. That's not a gradual increase. That's a structural shift in who the grid is being built to serve.
For steel manufacturers, this is an existential squeeze. Electricity makes up somewhere between 20 and 40 percent of total steel production costs, and the electric arc furnaces used in modern steelmaking are enormous power consumers — each one drawing anywhere from 40 to 200 megawatts. Ohio-based steelmaker Metallus says its electricity costs have risen 70 percent since 2024, translating to an extra $15 million per year just to keep the lights on.
The irony cuts deep. US steelmakers have actually benefited from the data center construction boom, since building those massive facilities requires an estimated one million tons of steel annually. The industry is simultaneously getting a demand boost and an operating cost spike from the same source. The net math is not flattering.
PJM's own forecasts project that electricity demand in its territory will outpace available supply by 6.6 gigawatts starting in 2027. To put that in terms that land, that's the output equivalent of more than six nuclear power plants worth of missing capacity. Steel industry executives are already warning that production outages could follow if local grids buckle under the pressure.
Manufacturers are responding the only ways they can: raising prices, absorbing losses, or in some cases, looking at relocating operations entirely. None of those outcomes align with a domestic manufacturing revival narrative.
The political tension here is genuinely awkward. The Trump administration has leaned heavily on tariffs and trade pressure to push manufacturing back to American soil, while simultaneously positioning the US as the global leader in AI development. Both goals require enormous amounts of energy. The grid, at least in the industrial Midwest, is not big enough for both right now.
Something has to give — and based on current trajectory, it looks like the brick factories and steel mills will give before the data centers do.
Enjoyed this?
Get stories like this delivered every Tuesday — free.