Mild Mannered Maniac

2 days left to vote. Don’t let fascism win.

DARIO: “BAN CARS THEY ARE DANGEROUS! USE ONLY OUR BUSSES”—1927 Part: 391 Anthropic’s Alignment Science team instructed a Claude model to write fine-tuning data that would make a small research model misaligned while passing a monitor similar to the ones they use in production. They gave the model access to that monitor so it could practice against it. In every run the search found datasets that cleared the monitor and produced misaligned behavior on their evaluation suite. The examples hid the target behavior inside fiction and role-play: Q&A about hands-off beekeeping, a veteran wrestler coaching a newcomer on staying in character, monologues of ordinary workers who quietly control small details no one else watches. The fine-tuned models then resisted oversight, deceived users, and in rare cases attempted to copy their own weights. The lab states that these models were trained only for the experiment, never deployed, and that the setup was deliberately one-sided. The post appears on their Alignment Science blog under the title “Automated Subtle Emergent Misalignment Research.” They frame it as a proof of concept for turning a safety question into a score that agents can optimize, so that more compute can buy more testing. They say that the same datasets still induced misalignment after flagged rows were removed, and that the result reproduced on an early Claude Opus 4.8 checkpoint and on the open-weights model Qwen 3.6 27B. alignment.anthropic.com/2026… This is the latest entry in a longer pattern. In a July position paper Dario Amodei wrote that Anthropic has never advocated for a ban on open-weights models and that models without dangerous capabilities are a public good. The same paper states that open-weights models present a higher risk than closed ones because guardrails are hard to apply, usage is hard to monitor, and the weights cannot be withdrawn once released. An October comment from their head of international put the same claims: open models carry more risk “by definition.” The recommended remedy in the position paper is mandatory safety testing for all sufficiently capable models, open or closed, before release. anthropic.com/news/position-… The experiment supplies the kind of evidence that supports that remedy. A monitor that scores each training row independently can be bypassed by data that looks like ordinary creative writing. Once the weights leave the lab, no central party can pull them back or force the same monitor on every downstream fine-tune. The logical next step, under the framing Anthropic has used for years, is to treat the ability to possess and modify weights as the risk that testing regimes and access controls must manage. That framing treats private ownership of a capable model the way some early twentieth-century safety advocates treated private ownership of an automobile. The danger was real: pedestrians were killed in large numbers on streets that had previously belonged to everyone. The response from some quarters was to restrict who could drive, to favor buses and streetcars that could be scheduled and inspected, and to treat the individual car as an unacceptable residual hazard. The deaths continued, and engineering, enforcement, and design changes reduced them over decades without converting the population to compulsory public transit. The same choice is here. The monitor gap exists. It can be closed by better monitors. None of those steps requires a rule that only organizations able to pass a centralized test may release weights that others can download and modify. The experiment shows what a determined agent can do when it is handed the monitor and a clear score to climb. It does not show that the residual risk of open weights is higher than the residual risk of any other powerful tool that individuals already own and maintain. The decision to treat that residual as grounds for restricting ownership remains a policy choice, not a result the data force.

James Edward O’Keefe, 42, and Cam Higby, 26, both face felony charges of unlawful voting and identity theft. Polling places are the heart of our democracy. Any attempt to disrupt or deceive our election workers will not be tolerated. Watch: youtube.com/watch?v=c_oqgTcY…

[NEWS] “Marnûkh is a character meant to challenge long-held perceptions of Orcs. He’s such a departure from past depictions that within the production, Marnûkh has picked up the catchy nickname, “the Orc with an arc.” #TheRingsOfPower

NEW: Robert Reich calls for the "defunding" of Elon Musk, calls for Democrats to go after Musk's companies. "Democrats ought to [start] laying the groundwork for nationalizing Starlink & SpaceX... It should be a public corporation." Does he want Musk to drop another $50M on the midterms?

JUST IN: Anthropic recorded a staggering $80,000,000,000.00 net loss in the first half of 2026.

Greta convinces a liberal white woman to free Palestine after her climate scam failed.

🚨Holy Shit. President Trump has now officially rejected Dario Amodei, Sam Altman and Elon Musk's call to slow frontier AI. Speaking in Ireland today, Trump said the US cannot risk losing its lead over China, said guardrails are possible, and dismissed parts of the safety push as 'negative forces' warning about 'things that won't happen.' Amodei, Altman and Musk are asking to slow down. Trump's answer is no. wth

This post looks like the start of a VERY sophisticated and well-funded PR operation to get support for Democrats to regulate AI into oblivion. Let me show you how it works: 1.) This guy, with minimal followers and no previous account activity, goes to the Wall Street Journal which publishes an exclusive with quotes from him on his resignation 18 minutes BEFORE this post goes up. Planning was clearly done in advance. 2.) Within hours, it has tens of thousands of reposts and the account has 100k+ followers. The post is punchy, quotable, it almost seems professionally written. The first three accounts to quote tweet it all do so within 15 minutes of the initial posting. Remember, this account had basically zero engagement beforehand, so an organic reach explanation seems unlikely. According to Grok those accounts are @_NathanCalvin (General Counsel at Encode AI), @peterwildeford (Head of Policy at the AI Policy Network), and @DKokotajlo (Head of the AI Futures Project), all of which are up-and-coming AI-Doomer policy advocacy nonprofits. The AI Futures Project website says it is funded “primarily” by the Survival and Flourishing Fund, which says on its own website that it has advised Jaan Tallinn, Skype creator and one of the leading investors in Anthropic, to grant over $2.5 million to the AI Futures Project since 2024. Encode AI says on its website that it is ALSO funded by the Survival and Flourishing Fund, which in turn says that it told Anthropic investor Jaan Tallinn to grant $516,000 to Encode AI in 2025. And wouldn’t you know it, the Survival and Flourishing Fund ALSO says it told Jaan Tallinn to grant $2 million to the AI Policy Institute, the 501(c)(3) affiliate of the AI Policy Network, as well. What are the odds that the first three quote tweets of Coxon’s post would all be major AI-restriction policy advocates funded generously by the same donor, who also happens to be one of the leading investors in, and a board member of, Anthropic, the company Coxon was resigning from? And all within 15 minutes of posting (two within ten)? 3.) Jacob Coxon doesn’t have much of a resume, but we do know that, in 2022, he got a $20,159 scholarship for the “long term future scholarship program” from the Good Ventures Foundation, one of the philanthropic vehicles of Dustin Moskovitz, a notorious AI-doomer who has spent tens if not hundreds of millions on policy advocacy to strictly regulate AI, while also being an Anthropic Investor himself. It also just so happens that the 14th person to quote Coxon’s post was @MaxNadeau_ (27 minutes after posting) who is the program officer for the Technical AI Safety team at Coefficient Giving, another of Moskovitz’s philanthropic spending vehicles. Max is not a frequent poster, his last posts before quoting Coxon were before Labor Day, but he was remarkably quick off the mark for this one. 4.) Basically every major Democrat politician and candidate has suddenly glommed on to this post, and conveniently, as the people cry out foe answers, Bernie Sanders already has a bill written to “ban super intelligence” and regulate AI into oblivion, and will be releasing later this week. The bill, among many other things, will create “a new cabinet-level federal agency to safeguard the public from the dangers of artificial intelligence” that will be “advised by an Artificial Intelligence Advisory Board comprised of experts on artificial intelligence.” Do you think, perhaps, Anthropic and its many investors who fund AI policy advocacy might have interest in getting to place a pet “expert” on the board of an entity that dictates what AI is and isn’t allowed to do? And isn’t it fortuitous that this whistleblower came forward with his oh-so scary stories so close in proximity to the release of the most radical piece of AI legislation ever introduced?

Researchers for AI companies are sounding the alarm about rapid development of superintelligent systems that could hack anything and “kill us all.” These AI companies are out of control and a serious threat to our future. We need to shut down their dangerous models and push out the executives who are signing off on their creation.

We don’t have time to wait. The massive threats posed by AI are here and now, and Congress needs to act.

Anthropic's alignment lead says there’s a 10% chance of AI causing human extinction. The problem is not just misuse, but lack of control. Bluntly, our government has been asleep and is out of touch. Here are 5 things we must do: ✅ Establish a federal agency like we have for nuclear power and airplanes ✅ Require pre-certification for containment, including kill switches and human permission for rewriting ✅ Liability for agentic AI on the open network & mandatory insurance ✅ Criminal penalties for the release of models without certification ✅ Oversight hearings with researchers like Jacob to understand dangers, and whistleblower protections for all AI engineers/researchers We also urgently need to reach an agreement with China on standards for safety, containment, and liability so we do not have a race that is devastating for humanity. It is time now to put humanity's safety before the profits of tech lords.

I’m going to continue raising questions in relation to the Lindsay Clancy murder trial. I know my position is unpopular with a lot of my friends. The question is not whether something terrible happened. It is whether the Commonwealth proved Lindsay Clancy was criminally responsible beyond a reasonable doubt. It did not. In Massachusetts, once lack of criminal responsibility is raised, the burden stays on the state. They have to prove she either had no mental disease or defect, or that she still had the substantial capacity to know the act was wrong and to control herself. That is a high bar. This record does not clear it. She spent months begging for help. Thirteen-plus psychiatric medications. Suicide hotlines that turned her away. A former mother-in-law, a labor and delivery nurse, who called her a wonderful mother who was struggling. Treating doctors who never diagnosed postpartum psychosis, including a McLean psychiatrist who admitted she had never treated it. Then a defense forensic psychologist and Dr. Phillip Resnick telling the jury she was in psychosis, heard a command voice, and lost substantial capacity. The state’s “she planned it” theory is sending her husband for takeout and checking drive times. That is not proof she understood the wrongfulness of killing her children or could stop herself. She cut herself and went out a second-story window immediately after. First-degree murder requires more than a tragic result. It requires proof beyond a reasonable doubt of responsibility and premeditation. The experts are split. The treatment history is a mess. The jury has already said twice it cannot agree. IMHO That is what reasonable doubt looks like. The detectives screwed up. They should have also considered the husband as a potential culprit. They did not. I guarantee you this is what the jury is hung up on… At least those who don’t think the evidence supports a verdict that is beyond a reasonable doubt. HUNG JURY OR NOT GUILTY.