[–] 13 points 3 weeks ago (2 children)

https://arstechnica.com/security/2026/10/the-internet-goes-on-high-alert-after-discovery-of-8-compromised-tls-certs/

It's hard to properly describe how big a fuckup this is. All I will say is that all the coverage I've seen is underselling the potential for harm. There's nothing you can do except update your shit, pray that everyone in the upstream supply chain of every service you use has updated their shit, and take comfort in the fact that there are no proven wild exploits out there (yet).

  • source
  • [–] 6 points 4 weeks ago*

    I'm not aware of any realistic way to vet published model weights like that.

    It works for some open-source software, because you can download the source and compile it yourself. But most of the "open" models don't publish their entire training set, or enough details on the training for it to be truly reproducible. DeepSeek doesn't publish their data sets, and they probably couldn't, legally, since they "collaborate with third-party data providers to obtain proprietary datasets". And even if they did, it would be prohibitively expensive to train it on your own, unlike simply compiling open-source code.

    The actual code for training LLMs is relatively trivial. The secret sauce is the training data and fine-tuning.

  • source
  • parent
  • context
  • [–] 9 points 4 weeks ago (5 children)

    It's still a risk. You can train an LLM to respond maliciously to a very narrow range of requests that would not be easy to spot in general testing. If you know how the LLM is used, you can target that very specifically. It doesn't have to be as blatant as "always deny the Tiananmen Square massacre ever happened".

    Given how thoughtlessly people use AI, and how opaque some of the systems are, I think this is a legitimate concern.

    And yes, every government outside America should ban the use of ChatGPT and Claude. Obviously.

  • source
  • parent
  • context
  • [–] 45 points 4 weeks ago (25 children)

    Honestly, reasonable.

    I mean, I sure don't trust OpenAI or Anthropic any more than I trust DeepSeek, but that's because I don't have legal control over them. The US government does, and it is correct to ban tools that are controlled by adversaries.

  • source
  • I've been lucky with my upgrade/build timings. Built my first PC right before the crypto madness, and did my last major upgrade right before the AI madness.

    I'm going to run my current system until it crumbles to dust. With any luck, the market will have settled a bit by then.

    As for projects, I have lots of old "useless" tech that I can work with. Like old laptops, phones, Raspberry Pis, even mobos and other spare PC parts. Building small is more interesting to me than building big.

  • source
  • [–] 1 point 1 month ago (1 child)

    You can strangle somebody with your arm.

    But can you strangle someone with a dead arm? It sounds very different than using a live arm. Without any muscle activation I think you'd be putting a lot more stress on weak points in the snake's body and it might just tear apart. But I have no first-hand experience strangling people with live or dead arms (or snakes), so I'm just playing armchair murderer here.

  • source
  • parent
  • context
  • view more: ‹ prev next ›