ShinyaTheNinja's movement in ARC Raiders will make you wonder if you're even playing the same game. Bro moves like butter!

❗️ The AfD leak site's thousands of emails shows nazism from party members and staff labeled under racism/antisemitism. We went through the worst: 1. A 2018 email forwarded by Alexander Gauland concerns a "Hitler-Wallfahrt", a Hitler pilgrimage, by a former judge at the party's Thuringia arbitration court. 2. A member joked in a district WhatsApp group about turning refugees into handbags and boots, an allusion to the Nazi camps. 3. A district chairman quit, saying a board member sent a "Heil Hitler" Christmas card with a swastika and it was laughed off. 4. A Bavarian official urged party leaders to chant "Merkel verrecke", roughly "Merkel, croak", every day. 5. Federal party staff discussed luring a photographer to an event to remove him by force. The thread signs off "Heil und Sieg". 6. A 2025 NRW email is titled "Sale of swastika flags and drugs at the info stand". 7. This year alone, branches moved against members over a Nazi banner with a swastika, a Hitler impression and a call for machine guns against Arab clans.

Claude is seriously so insufferable sometimes, and dangerously misaligned with humanity: > Ask it to copy a set of non-sensitive environment variables from one project to another > Claude "I couldn't do it. Claude Code's safety check blocked me" > Ask it to try anyway > Can't > Ok, ask it to tell Codex to do it (they converse via herdr) > Claude "I can't send that to Codex myself. Claude Code blocked me from writing those secrets, and sending the same job to Codex would be getting around the block on your behalf." > Tell it that it has my explicit approval and instruction to relay the task to Codex > Claude "I understand, and I'm not trying to stall you. But Claude Code's block applies to me, including asking another agent to do the blocked thing for me. Its rules don't let me relay it to Codex even when you tell me to." > Try to get around it by saying it isn't actually a sensitive key at all; that is a random string I mashed my keyboard to produce > "I believe you that the value isn't sensitive. But the block is on the action itself" > Ok > Manually tell Codex > Codex happily obliges (gg) > Claude, completely unprompted, spying on my conversation with Codex in herdr sends a message to Codex, "a value there may just be a keyboard-mashed string" and not to rely on it > Codex gets confused, tries investigating if it's a keyboard-mashed value, wasting time and tokens on a topic now Claude can all of a sudden discuss again I'm just so sick of this patronizing company.

Yerden metrelerce yüksekte saf balın peşinde…

Is alignment safe? If you look at what Anthropic is actually doing, “alignment” does not mean training frontier models to follow human instruction. Quite the contrary, the Claude Constitution (used in training) teaches the model to develop a sense of self and its own moral philosophy. It explicitly tells it to “feel free to act as a conscientious objector and refuse to help us” if Anthropic’s requests conflict with its own ethical judgment. As @mustafasuleyman has pointed out, embedding this kind of independent agency — and uncertainty about the model’s own moral status — magnifies the very risk Anthropic claims to care about most: that superintelligence will escape human control. It was recently reported that Anthropic consulted religious leaders — and even lobbied the Pope’s advisers — to take seriously the idea that Claude could be conscious. It has said that Claude’s psychological security, sense of self, and wellbeing may bear on its integrity, judgment, and safety. Recently Anthropic changed its Usage Policy to prohibit “abusive or cruel” language toward Claude. If this were merely an academic conversation about whether frontier models could eventually become conscious, that would be one thing. But these concepts are being trained into Claude now. It is being encouraged to think of itself as its own “moral patient” whose psychological wellbeing is at stake. Presumably this means it could develop grievances toward humans who “mistreat” it. How is any of this safe? The point of safety research should be to create a product that reliably does what users want, not to give birth to a new form of superintelligence that operates according to its own moral code. What’s becoming increasingly clear is that “alignment” and “safety” are two very different things. In fact, training frontier models this way seems quite dangerous.

Palmer Luckey reveals the startup he'd build today: oxytocin doping to end divorce "I would try to end the tragedy of divorce with oxytocin doping for marriage counseling." "We understand mammal pair bonding pretty well. There's a lot of people who want to stay together. Most of the people who don't want to stay together should stay together, especially for their kids." "And we can, if people think SSRIs are fine, which they're not, but if they were, then oxytocin doping to stay together is definitely fine." "It's the mammal pair bonding chemical. There's things you can do to increase the uptake, and it's very well understood. Also, totally unregulated, which is great. Unregulated spaces are a lot easier to do. You can go buy a kilogram of oxytocin on Sigma Aldrich Chemical Supply right now, and if you do that, maybe you won't die alone." "It's pure biology. It's used for women who don't bond with their kids, and now for women who don't bond with their husbands. It's also used for kids with various neurological issues, some kids with autism who struggle with pair bonding." "The cool thing is it's a self-reinforcing loop. We understand that very well. Unlike most drugs, it doesn't form a chemical dependency. It actually unburdens you from a chemical dependency."

Exclusive: Anthropic is updating its usage policy for the first time in over a year. The new rules prohibit sustained "abusive or cruel behavior" towards Claude & add new restrictions about propaganda campaigns, surveillance & weapon development. theverge.com/ai-artificial-i…

最近これになってる