The Post wanted you to think Sam Altman is fine with killing babies. Here’s how they built it.

“Nutty professor” is an insult, not an argument.

The name-calling starts before the explanation. Later, “doomsday cult” and “messiah complex” do the same kind of work: they give you a reason to dismiss the people before weighing their case.

None of those labels tells you whether Singer’s argument is sound, whether an AI forecast is credible, or what a proposed safety rule would do.

Singer’s case for euthanasia comes with conditions the opening leaves out.

Singer does not give blanket approval to killing disabled babies. His examples concern particular decisions and circumstances. Some involve suffering; the hemophilia example later in the article goes further, involving a baby who could have a happy life.

The sexual examples also make assumptions: an animal treated as willing, or permission given before a person’s death. Those assumptions are part of what he is defending. They are not proof that real cases meet them.

Agreeing about charity doesn’t mean agreeing about everything.

Effective altruism starts with a practical question: where can your money, time and effort do the most good? People can share that goal without agreeing on every answer, let alone every opinion Singer holds.

Amodei wrote about effective giving in 2010. That connection is real. The NY Post offers no statement from him or Altman agreeing with Singer about killing newborns or sex with animals.

They call it a power grab. They don’t show one.

The charge seems to be: frighten people about AI, win support for strict rules, then use those rules to keep competitors out. “Tightening ranks” appears to mean the companies acting together.

The NY Post names no agreement and shows no evidence that the warnings are being used for that purpose. A rule might help large firms. That is different from proving their safety concerns are a cover for gaining power.

Right and wrong are the whole point of Singer’s work.

His test is who suffers, who benefits, and how much. He does not think following a rule automatically makes an action right.

Take lying. Suppose someone asks where to find a person they intend to kill. In that example, telling the truth could be the terrible choice. The harm matters more than the rule against lying.

That is a way to judge right and wrong, not permission to do whatever you like. The example is ours; the approach is the one Singer describes.

The line left out: killing a baby is almost always terrible.

Singer says this in his own FAQ, giving great weight to the harm done to parents who love their child. He allows exceptions decided by parents and doctors.

He also uses “person” to mean someone who can imagine their own future, a capacity newborns do not yet have in his account. His exceptions go beyond unbearable pain. The hemophilia example concerns a baby whose life could be happy.

He’s comparing what death takes away, not which species is “better.”

Singer asks what an individual can feel, understand and lose by dying. On his account, being human does not automatically settle the answer.

That really can put an adult animal’s continued life ahead of a newborn’s. The comparison is not invented. But it is about the loss to each individual, not a claim that the baby has no value.

The interview exists. The 2023 share is a separate claim.

Published interview excerpts show Singer discussing sexual acts under stated conditions. They do not verify what he posted or shared in 2023.

The NY Post reports that later share. We have not independently checked the social-media post, so we are not treating its date or contents as confirmed.

The Nazis ordered killings. Singer puts the decision with parents and doctors.

The Nazi state organized the murder of disabled children and adults in the name of racial “hygiene,” often deceiving their families. Singer describes decisions about individual newborns made by parents and doctors.

Who decides, whose interests count, and why the life is ended are central differences. Walter’s comparison does not make them disappear. The criticism that Singer’s approach devalues disabled lives still needs to be argued on its own terms.

He calls the child’s life worth living. Then asks whether it can be replaced.

Singer starts with a baby with hemophilia who could have a happy life. If the question is what is best for people already alive, he says killing the child would be wrong.

Then he compares two futures. In one, that child lives. In the other, the child is killed and the parents have a happier baby who would not otherwise be born. If nobody else is harmed, he says a rule favoring the most happiness overall would choose the second future.

That is an argument for killing in that case. It is not the blanket claim that children with hemophilia “don’t deserve to live.”

The movement came first. Singer’s TED talk helped spread it.

Giving What We Can began in 2009. The career-advice group 80,000 Hours and the Centre for Effective Altruism followed in 2011. Singer gave the talk in 2013.

Calling him a “father” of effective altruism makes sense as a description of his influence. It does not make every later idea his. The Centre’s history also describes the name as a group choice, not one person’s invention.

The point is to do more good, not make more spreadsheets.

Two charities tackle the same problem. One helps many more people with the same money. Effective altruism asks you to consider that difference before giving.

The numbers help compare the options. Helping is the goal. Whether the numbers leave something important out is a question about the calculation, not a different purpose for the movement.

Some earn more so they can give more. Others do the work directly.

“Earning to give” means choosing a better-paid job so you can donate a substantial share of the income. It is one way to contribute, alongside research, direct work and other careers.

It is not an instruction for everyone to get rich. And the published definition requires actually giving money away, not simply promising to do it later.

The idea is to help where it matters most, not pick the most distant charity.

Singer argues that distance alone should not decide whose suffering matters. The question is how much good a donation can do, not whether its destination sounds familiar.

Walter’s examples raise a real choice: favor a local cause, or help more people elsewhere? That is the disagreement, not a preference for faraway places for their own sake.

Singer has said the opposite: moral truths do exist.

In a 2019 interview, Singer said he had come to believe that moral truths do not depend simply on what people happen to prefer. That directly challenges the earlier claim that he has no right and wrong.

It does not tell us what every effective altruist believes. Thayer’s claim about what draws people to the movement would need evidence about those people.

He said he wanted to give. That doesn’t prove what he really wanted.

Bankman-Fried presented getting rich as a way to help others. The NY Post uses that account as an example of “earning to give.”

His stated goal belongs in the story. But “started out wanting to help the world” goes further: it treats that public account as proof of his private motives.

Future people matter more in this argument. The NY Post says less.

Longtermism starts with the idea that people who may live in the future matter too. Some versions give them enormous weight because there could be so many of them.

The NY Post says those future lives become insignificant. That is backwards. The disputed question is whether possible benefits far in the future should outweigh help for people alive now.

AI risk wasn’t a new cause after COVID. It was on the list in 2016.

The NY Post moves from animal welfare to COVID, then says the movement has “now moved on” to AI. That sequence leaves out years of overlap.

A December 2016 donation guide already recommended AI-risk work alongside pandemic preparedness, animal welfare and global health. Priorities could have shifted later. The interest in AI did not start after COVID.

AI researchers do publish timelines. AI 2027 is one.

First published in April 2025, AI 2027 sets out dates and a detailed account of how its authors think AI could become dangerous.

You do not have to believe the forecast. But it exists. That is enough to disprove “never.” It does not tell us how often the wider field makes forecasts, or how reliable they are.

Which rule would shut out small companies? The article doesn’t name one.

Walter suspects the biggest AI companies want rules that make it harder for others to compete. The NY Post does not examine a particular rule, who would have to follow it, or what it would cost.

We get his suspicion about the companies, not evidence of what a rule would do. Those are different claims.

Amodei wants outsiders checking his company, too.

His proposal gives independent evaluators access inside AI labs, including Anthropic. They would be able to publish unfavorable findings, with limited exceptions for protected information.

Those details matter to the power-grab claim. They do not prove the plan will work or rule out costs for competitors. They do give readers an actual proposal to examine.

Singer rejects the “AI doomer” label. The NY Post includes his reply.

He says he has never belonged to such a movement. That response appears at the end of the article.

He has discussed AI and other long-term risks, including in a 2019 interview. Talking about a risk is not the same as backing every prediction or policy made in its name.