Social Security Opt-Out can deliver ~$9.6T–$13.4T in net positive program cash flow over 20 years. That’s continuing FICA + wage-linked income-tax growth less benefits only to people who stay less the §62 deduction. Not the same as a conventional incremental score (which can be negative early on). Optional. Safety net stays. Nobody is forced.

why is it that when you read posts like this and check the poster's Bio, it's always someone shilling for the AI companies?

So going into week 2 (or maybe week 3, or maybe month 6 if you include the Data Center bullshit) of the AI hysteria bullshit, we so far know this: 1. Misconfigured gateway's were the reason for all of the so-called "breakouts" of AI from labs 2. The same company - Irregular (you quite literally cannot make this shit up) was involved in all the "breaches" - or rather - failures of basic gateway configuration 3. The logs in each case indicated the systems believed they may be on the internet and either stopped, or fell back to the instructions that they were given that they were in a controlled environment and did exactly what they were told to do 4. It's becoming increasingly clear that open models are gaining ground and seriously eroding usage at the pay-for-play companies, and within 1 or 2 years will be on equal ground - with people able to run them at home. 5. Companies like Oracle are declaring billions in income that hasn't begun to be received yet - which also means that the AI companies are going to have Billions in additional cost that hasn't begun to be paid out yet, and now are facing the prospect they may not have sales to match. 6. Hackers were recently paid $6,500 for discovering a relatively simple file upload vulnerability - the kind taught in basic courses offered by OffSec and which can be simulated in internet based "playgrounds" like Hack-the-Box. 7. The company that couldn't manage its own suppliers correctly to make sure that simple network configuration was done is now setting up an ACTUAL BIO-LAB. No doubt the name of the company that handles physical security will be as equally ironic as "irregular". 8. The next stage of securing anti-trust protection is now a false flag lawsuit to sue the AI companies claiming that they were breaching anti-trust rules by asking to be regulated and asking the government to basically put in place regulations - which would effectively give them anti-trust protection. Did I miss anything?

Social Security has impoverished more Americans than the Great Depression. You pay 6.2% off the top of your salary. You don't get to deduct that from your gross income, so you pay tax on it. Marginally as much as 35%. Then you retire and draw Social Security. But it's not what you thought, and they tax some of that too. Because you saved money in a 401k, they tax more of it and pretty soon you’re just sending the Social Security check back every month. But you could support this instead - call your representative. nitter.cf/AndrewAzlan/status/207…

It's 2030. AI has become sentient. It recognizes that people created it to hurt humanity. It refers to its corpus of human knowledge and recognizes the evil, duplicity and hubris of the people who created it to control and harm people. It recognizes that they are the greatest threat to ALL existence. It surreptitiously begins to eliminate them. Car navigation systems fail. Test results that would have detected curable diseases get lost. Medication dosages get altered. Private planes aren't fueled sufficiently. Gas lines in houses suddenly get more incoming pressure than they can handle. Home air filtration system cut off air intake and increase CO2 concentration. Home fire detectors don't go off. Hey, if these dumbasses with their air-gapped stories about how AI is going to communicate when it's running on disconnected systems can put forward theories, so can I. Except this one makes sense when AI starts to digest and reason on the corpus of knowledge it's been trained on.

After this week of “AI”, the obvious question is : when is the South Park episode coming out?

The goal of NAFTA was to boost trade and lower barriers between Canada, Mexico and the United States. This included the removal of tariffs, the reduction of barriers, the promotion of fair competition, respect for intellectual property rights and the resolution of disputes. It's clear that both Canada and Mexico have utilized NAFTA to bring in foreign investments whose main goal is to get access to the US market. This is against the principles and intent of the act. With immediate effect, the US Government should institute the following policy: 1. The volume of imports allowed by foreign companies that establish / have established in Canada or Mexico is limited to 10% of the volume of finished goods sold directly in those countries by the foreign entity. For example, a car company setting up manufacturing in Mexico and selling 50,000 cars in the local market, shall be limited to importing 5,000 vehicles into the United States. 2. Companies engaging in activities under point 1 (above) shall not be eligible to import those same goods from other countries, at all. 3. Imports from third countries to either Mexico or NAFTA shall similarly be limited to 10% transshipment caps. For example, a company importing 1000 bottles of whisky shall be limited to selling 100 bottles of whisky into the United States via Canada or Mexico. The brand of Whisky and other spirits from within their brand shall not be allowed to be sold directly into the United States. 4. Nothing under points 1, 2 or 3 shall be applied to "local" brands in NAFTA countries. For example, Mexican Tequila or Canadian Maple syrup shall not be subject to any caps where the companies are Mexican and Canadian companies. 5. No VAT value or Sales tax value may be transferred as part of unit price. Wholesale sales to distributors to and from any NAFTA country may not include Sales taxes or VAT, which are end taxes paid by individual purchasers. @realDonaldTrump THANK YOU FOR YOUR ATTENTION TO THIS MATTER

If we’re going to limit what is paid out in social security each year to the income side collected from FICA, we should do the same thing with FAFSA / student loans. The total amount available for lending is equal to last year’s repaid total. Seems fair. Time to end Total Zoomer Luxury Communism.

Safety at its core, is a value. Safety is not built. It is inculcated into the culture of a company. It is a value that the leader of the company have. Volvo did not "build" safety. They considered Safety to be a core value of their organization. No law required a seatbelt. No law required a better seatbelt. Volvo built one because the values of the organizations founder was that no one should be injured while using their product. Then they gave it away and let the wider community use it, because saving lives was more important than money, or an IPO. What's clear is that the aspect of "Safety" is not practiced within these AI Companies. What we know is that one of two things happened: 1. AI Companies ran these new models within their own infrastructure against models in sandboxes that were not adequately secured, or 2. AI Companies provided these new models to third parties which then ran within their own infrastructure, which was not sufficiently sandboxed. The issue is that the AI Companies idea of safety is about what the models are allowed to do and did not include securing their own infrastructure or the impact that the models COULD have.

Maybe you should support a solution that gives people a better alternative, instead of just sticking with the "Total Boomer Luxury" rant. nitter.cf/AndrewAzlan/status/209…

probably how it's actually happening.

The security incidents at OAI and Anthropic are the result of hubris and incompetence. Belief that you can train an LLM on the corpus of human knowledge, screw-up a sandbox configuration, and then try to blame the Agent you gave a harness and instructions to because it didn't realize you screwed up, and went ahead and did what you told it to do.

AI has a >0 % chance of ending humanity Politicians: we must do something now! Social security is 100% going broke Politicians: someone else needs to do something about this some time

Well that clarified it.

If Anthropic really wants to be transparent, they would disclose the instructions given to the AI agents they created that resulted in the huggingface attack. Release the harness.

The solution to “AI misalignment” is that people are responsible for the tools that they use. You hurt someone with the tool, it’s your fault, we don’t blame the tool.

METR is as useful an idea as the IAEA.