Most of this is just marketing crap from Anthropic.

Finding vulnerabilities in code and generating complex, multistep exploits with publicly available models is possible now. This biggest hurdles now is setting correct context and actually knowing what to look for. Any “guardrails” for this behavior are easily bypassed by framing the detection and exploit generation as a valid dev style question in the most difficult of situations.

They likely just trained a model without guardrails in this case.

What they are doing here is over-hyping a problem and framing it like they are the only ones with a solution. LLM security issues are more in-focus now that companies have dumped a ton of resources into building AI systems they don’t really understand.

You should probably have your bot stop at a word count, not a letter count as to avoid unfortunate racial slang.

It’s probably just a rogue hemorrhoid that will just add more stress. Sorry.

Nobody said anything about taking a shower.

Companies these days do what is right for their shareholders and if Claude makes money, or appears to make money, then the shareholders are happy.

Who votes for the Supreme Leader of North Korea? The prime minister?

A response that is actually in context: Considering how expensive it is to build a fab for a component as critical and delicate as RAM, there is incentive to perform proper QA for products released under the original brand. Having a fab fail because of reputation is not likely an option. Rebranding wouldn’t help as modules can be de-capped and the source vendor could still be identified.

The success or failure of this vendor is going to be how well they physically control their bottom tier bins and ensure that any waste product doesn’t get funneled back into the supply chain. With China specifically, it seems the incentives are much higher for that behavior. Again, if you doubt that, I can point you in the direction of thousands of bunk components.

As with any company that is state owned or state backed, the potential security risk is much higher. I am not just pointing directly at one country in this case. Some governments may pose higher risks than others though. (From a security perspective, you would want trojaned components to be as reliable as possible, TBH.)

Unless there is strict 3rd party (out of country) quality control or there is financial motivation for proper QA, Chinese electronics are usually trash. The market is flooded with cheap Chinese silicon fakes, which has caused significant price increases to verify legitimate parts. If its not “original” pirated silicon that is the issue, it’s filed off package marking with a shitty re-badge.

You can keep barking that nationalist bullshit, but it doesn’t change the fact that I have to rebuild any equipment that I need at a slight discount and don’t want it to kill me because of a 2 cent savings on a missing ground.

Unlike you, I don’t give a flying fuck about talking shit about another country. It’s the electronics that matter to me, and if you haven’t seen the absolute shit show that is Alibaba, you have your head so deep in the sand you are never going to experience that sweet smell of burning, pirated XT-60 connectors.

I have delt with so many fake parts smuggled into legit supply chains it would make your head spin. This isn’t a “buyer beware” issue: it’s complete lack of respect for anyone else further down the supply chain.

At least stay on-point if you are trying to defend something, FFS.

From the research papers I have read, psychedelics introduce a degree of neuroplasticity that allows psychotherapy to more effective. (I cannot speak to what types of psychotherapy would be more effective than another as I do not know or understand the differences.)

I attribute my use of psychedelics to helping me through my alcohol addiction. While it wasn’t guided therapy, it was still very controlled and allowed me to “rewrite” how I interpreted feelings and how I handled a variety of different situations. My hallucinations allowed my feelings to become more tangible and physical. I felt I had the opportunity to think differently about difficult parts of my life.

I suspect proper guidance is similar to what I did to myself: Have a person describe situations and the therapist proposes different ways to interpret those situations. The brain is able to physically form new pathways and sidestep old behaviors.

You could probably look at RAM and GPU prices and come to the same conclusion much faster.

69% of developers report losing 8+ hours weekly to inefficiencies—20% of their time (Atlassian, 2024)

Yep. If there is any company that knows what it takes to drive inefficiencies, it’s Atlassian. (Fuck them and their software.)

These are actually real, for those who care: https://www.accuweather.com/en/weather-news/mist-cannon-leaves-rainbow-trail-floating-over-city-avenue/1195796

That’s kind of the point. You want the sheep to think they are burning down society for some do-good ideology while the politicians sidestep the government into a into dictatorship or other kind of single party authoritarian system. Regardless, the politicians of the new government will be just fine.

No. It might be in a politicians best interest to ensure a population stays dumb and then ensure those dumb people raise even dumber children.

Yeah, I tried to steer clear commenting on whatever people choose to follow politically or socially. It is a highly curated echo-chamber, is my point. (Most social media rabbit holes are echo-chambers, but ml wins awards for that.)

Just ignore ml. If you got banned for not knowing their version of whatever, you probably are just going to get banned again later.

For once, I am not knocking on ml for being ml, it’s just a very odd instance when it comes to social interaction. You must subscribe to the hive mind 100% … or else.

The highest estimate I could find was 167k KIA for Ukraine which aligns with other estimates of 500k total casualties. 1:3 is a fairly standard KIA/casualties estimate for many conflicts.

Ukraine is notoriously tight-lipped about their casualty counts though. As with any conflict, expect inflated/deflated estimates depending on who is stating them.

It’s within reason for a 1:3 (up to about 1:5) Ukranian/Russian casualty comparison, given the difference in tactics between the two sides. (Agree or not, Russia historically uses scorched earth and mass to fight wars.)

Unfortunately, you can get high/low estimates from all over the place. The actual numbers may never be known or not known for years after the conflict is over. Orgs that do verified tracking are generally very low with their numbers.

That is a shit way to normalize numbers like that, but averaging out all “officially” reported data and jamming them against rough ratios is all we really have.

(Edit: A recent escalation in Russian casualties is plausible as their access to new/replacement equipment has been reduced significantly. Dig a little into sources that scour satellite photos of tank depots or how sanctions are affecting manufacturing for more info. It’s a deep rabbit hole, but data is out there.)

You can interpret anything how you choose, kind of like we have to do with your grammar.

You are missing my point, but I also wasn’t clear enough. In proper context, we are saying the same thing.

I worded that sentence carefully, as to your point, I don’t actually want to tell people to go to Reddit. However, each platform is unique in its own way. If someone wants the Reddit experience, that is the only place they are going to find it. Reddit content is generally curated algorithmically while Lemmy content is not. It’s could be the same articles on the same day, but two different experiences.

OP was referring to reposting content for someone who seemed to be looking for the same volume of content that is on Reddit that is heavily sorted, unless I missed something. I was just saying that this platform doesn’t really support that kind of thing in a constructive way. The articles and the presentation combined make the platform “content”.