Slashdot title: "freeze your call if someone starts undressing"
iOS Warning: "paused because you may be showing something sensitive"

Those are two very different statements. Meaning this system doesn't stop incoming content, it just gives you a speedbump when giving out nudes.

And where's the test data set? What skin tones does it work on? What body shapes? I assume this news would be bigger if Apple finally got all that right rather than all the other nudity detectors which get so many things wrong. If I wanted to trust this to protect my kids (0) then I'd want to know the edge cases it misses and info about false positives. Further this is a child protection service not an adult blocking service. Was training it on developed adults good enough to protect undeveloped kids? Or is it just checking if a % of screen matches an estimated skin tone?

Aren't LLM inputs the same as neural network imports? Meaning a bunch of nodes receiving character/token/numeric input? In that case there's no way to isolate data vs commands unless you give every input node an additional node that toggles on/off if the related node is being given prompt or data input. Without training it like that injection attacks will always be possible.

Pattern matching is AI...

You're on a tech site. The tech definition of AI is a system that mimics what an intelligent creature would do. The "artificial" refers to "intelligence" meaning "fake intelligence". It doesn't mean "man-made".

The media definition of AI is people made out of electronics instead of cells. Such a robot would be beyond the tech definition of AI since it would have real intelligence. If LLMs had real intelligence they would no longer be AIs.

Trump is trans? Is that why he wears so much makeup?

There are a few extra things that pass from mother to baby in addition to DNA and one of those is intestinal bacteria. Use of antibiotics changes your intestinal bacteria in detectable ways (not sure if changes breast milk too). I don't recall if it's the direct bacteria change or related epigenetic changes that get passed along to the baby in this case. Either way, it takes roughly three generations for those chances to fully disappear to get back to a 'default human' instead of one customized for the environment your parents and grandparents lived in.

It's certainly better to take antibiotics compared to dying, but there are long term changes in doing so. Do those changes matter? Who knows. Finding fully healthy, untainted humans to study is impossible now. No where on Earth is pure. Though antibiotic resistance is a real problem so people should still avoid unnecessary antibiotic use.

The more important generational thing to pay attention to is hormone disruptors. Those have far more noticeable effects on your baby than antibiotic use. If you plan on having multiple babies, keep milking yourself after you stop breastfeeding your first baby. A lot of disruptive things end up being detoxed through your breast milk so extending that period will reduce the load for your next babies. Best if you can do it before your first baby, but that takes significant effort.

Every generation does that.

That's because they don't. Survey based studies especially so and then their little biases get amplified when collected into meta-studies.

For processed meat vs other food studies, I haven't head of any research that did it properly. As one example out of many, how do you eat a hot dog? How do most people eat hot dogs? Do you microwave it, cut it with a fork, then eat it piece by piece? Do you grill it until it's half burnt, wrap it in a bun then smother it with ketchup and mustard? In a food study, both ways of eating a hot dog count as eating processed meat yet the wrap and dressings people use often outweigh the actual meat they're eating. 1 hot dog = 120 cal (w/1g carbs). 1 bun + 2 servings of ketchup = 150 cal (w/31g carbs). The two are not equivalent yet food studies treat them as identical.

Then lets use common sense. How does meat lead to type 2 diabetes? The pathway for eating carbs leading to diabetes is easy to understand. What is that pathway for meat? All these questionnaire studies claiming meat is bad yet not any of them show how the meat causes damage. There's a ton of other research directly looking at how carbs damage your body and blood vessels. There aren't equivalent studies on meat because at the cellular level meat isn't the cause of that damage.

However the slight increase risk of colorectal cancer for processed meat is true due to the unnecessary use of nitrates. That bit of science is well documented too. Also documented is the increase of bowel issues due to fiber intake which dwarfs that cancer risk. As always, the side with the biggest marketing budget wins the public opinion.

If you want a better research group, look at the people eating the carnivore diet and compare all their health problems to other diets. That'll give you a better comparison with the problems of eating significant amounts of meat.

That research says:

Meaning the blended had more fiber than the whole food since it chopped up the seeds we normally don't break. The research would have been better if it just looked at apples without seeds.

But the parent claimed juicing fruit is better than eating candy and that's simply wrong. Juicing removes most fiber and so the resulting blood sugar levels and then insulin spike is far higher than from eating candy. Plus, almost no one juices one fruit. It's more like 3 apples as a base plus other fruit. That's at a minimum of 60g of sugar whereas coke has 39g. The beneficial things in fruit don't outweigh their direct sugar content.

I used to juice and I juiced 'healthy' meaning vegetables with an eye on their sugar content. It. Tastes. Horrible. If your juice tastes good, it's not the health drink you think it is.

> Placing an arbitrary work from office policy in place doesn't do that

The standard practice is to give exceptions to people you don't want to lose. Then a couple months later you might retract the policy in order to entice new hires.

One of the initial things Palemoon did was strip out a bunch of accessibility features because they didn't want to bother supporting them. I've never looked at them again after that. Kudos if they're actually trying to do better now, but the way they started was doomed.

Other industries advertise their quantity discounts and those are normally real discounts since bundling higher quantities together has cheaper per-unit processing/shipping. The airlines are being sneaky about these price chances and there's no inherent discount in multiple people flying together. The backend processing of one person vs multiple people is the same. Now maybe if they were advertising it as a bundled discount for groups, but they're not. This practice isn't being advertised, it's not a promotion. Thus it's fine to flame them for their greedy money grab. All their actions point to it being a money grab.

And you don't even need to have committed a crime to have a criminal record. Ever been finger printed (like as a kid in case you were ever kidnapped or fell into a monster infected quicksand pit with your detached hand sitting on top)? You have a criminal record. Ever got pulled over? Criminal record. Ever had someone make a random complaint about you to the police, regardless of if they every contacted you about it or not? Criminal record.

A criminal record only means the police have a file about you, meaning they interacted with you more than a "Hi" as you walked by. That what the media means when they say "criminal record".

That would be horrible. It takes resources to digest something. If you create something the body craves, people with significantly overeat it trying to meet that craving yet failing as it isn't proving the expected nutrients. Worse, it takes more nutrients to digest that empty food so you'd simply get worse and worse eating that fake food.

Actually we already have that. It's called flavoring. That's why your beef soup tastes like beef and why people eat half a bag of Doritos at once (umami and salty thus mimicking meat yet without the nutrients your body expects from meat).

If you want endlessly eat whatever without getting fat, you have to stop swallowing what you chew.

Fat that isn't digested by your bile ends up blasting out your ass. If you're eating a significant amount of fat and you haven't given your bile production the time to ramp up, you're not storing the vast majority of those calories. Similar to how your urine will start getting sweet when you gain insulin issues. Anything above your max storage rate isn't kept. Though that storage rate is dynamic.

Granted they'd still be counted as calories out even though you didn't burn them. CICO is very hard to measure.

Obesity has spread far too quickly for genetics to be it's main cause.

Genetic studies are often biased towards genetic causes. Many overlook lifestyle. It's completely reasonable that someone who grew up with parents providing them a fattening diet maintained that lifestyle throughout adulthood and someone with parents using a healthier diet maintained that way of living. Genetics might not have anything to do with that yet so many claim that it's all genetics.

Until they can point to specific genes as well as what triggers those genes to become active, genetic claims are usually suspect.