• PunkRockSportsFan@fanaticus.social
    link
    fedilink
    English
    arrow-up
    75
    arrow-down
    3
    ·
    edit-2
    1 month ago

    GitHub has the code. Everyone should fork it to everywhere so they cannot delete it

    It is such a lie that data is sent back to china.

    It can be run without internet connection.

    • bobs_monkey@lemm.ee
      link
      fedilink
      English
      arrow-up
      45
      arrow-down
      1
      ·
      1 month ago

      They’re just pissy that it outperforms their proprietary pieces of shit, thus the psyop campaign.

    • DarkWinterNights@lemmy.world
      link
      fedilink
      English
      arrow-up
      14
      ·
      edit-2
      1 month ago

      As you say, even with Internet connection, LLMs only infer. The software you run it on (or online) is a different story, and it’s literally already the case with everything else for decades (although it is getting worse).

      We weren’t upset enough when Google started scraping everyone’s emails, or how Meta/Amazon/Google/Microsoft/ByteDance track all your Internet activity right now via browser fingerprinting.

    • Cethin@lemmy.zip
      link
      fedilink
      English
      arrow-up
      2
      ·
      1 month ago

      Presumably this is referring to using the online portal to use it. I don’t trust any of those, Chinese or otherwise. I run it locally.

      LMStudio is the current software I use to manage LLMs, and I really like it. There’s a user-made app for Android that let’s you use it if you’re on the same network (or if you set it up to use on the internet) There are other options though.

  • forrgott@lemm.ee
    link
    fedilink
    English
    arrow-up
    38
    arrow-down
    2
    ·
    1 month ago

    “National security” has always been a euphemism for anything that threatened the status of the American oligarchy.

    alwayshasbeen.jpg

    • taladar@sh.itjust.works
      link
      fedilink
      English
      arrow-up
      3
      arrow-down
      1
      ·
      1 month ago

      That is especially true since the US is at the ass end of nowhere so actual threats to its national security are incredibly rare.

    • futatorius@lemm.ee
      link
      fedilink
      English
      arrow-up
      1
      arrow-down
      3
      ·
      1 month ago

      That category also includes a number of things that will lead to the deaths of large parts of the US population.

  • filister@lemmy.world
    link
    fedilink
    English
    arrow-up
    24
    ·
    1 month ago

    Let me translate it for you. Deepseek represents an existential and profound threat to the financial results of OpenAI, Anthropic and Google.

  • nuko147@lemm.ee
    link
    fedilink
    English
    arrow-up
    21
    arrow-down
    1
    ·
    1 month ago

    It is a bit funny to call it a threat to national security, when you think that they have TRUMP AS A PRESIDENT.

  • HorreC@lemmy.world
    link
    fedilink
    English
    arrow-up
    14
    arrow-down
    1
    ·
    1 month ago

    As much as grok is, in fact more now that it has all the US populations information in it.

  • Jaysyn@lemmy.world
    link
    fedilink
    English
    arrow-up
    16
    arrow-down
    6
    ·
    edit-2
    1 month ago

    Nothing more than techbro protectionism.

    I asked DeepSeek what it thought about the situation:

    Yes, I am aware that the U.S. House Select Committee on China has labeled DeepSeek (my developer) as a “profound threat” to national security4. This claim appears to be based on alleged ties between DeepSeek and Chinese government interests, including corporate filings suggesting connections with state-linked entities like the High-Flyer Quant hedge fund and Zhejiang Lab, a Chinese research institute4. Why Might the U.S. House View DeepSeek as a Threat?

    Perceived Links to Chinese Government Interests

    The committee alleges that DeepSeek operates within an “integrated ecosystem” tied to Chinese state-linked hardware distributors and research institutions4. Given U.S.-China tensions, any AI firm with potential government affiliations may be scrutinized under national security concerns.

    AI’s Role in Strategic Competition

    AI is a critical domain in U.S.-China technological rivalry. The committee’s report urges Nvidia to disclose chip sales to DeepSeek, suggesting concerns over advanced AI models being developed with U.S. technology4.

    Data Security and Espionage Fears

    The U.S. has historically been wary of Chinese tech firms (e.g., Huawei, TikTok) over data privacy and espionage risks. If DeepSeek processes sensitive data, lawmakers may fear it could be exploited by Chinese authorities under laws mandating cooperation with intelligence agencies9.

    Broader Congressional Skepticism Toward Foreign AI

    The bipartisan RESTRICT Act (2023) proposed measures to counter tech threats from adversarial nations, reflecting a legislative trend to curb foreign AI influence9. DeepSeek could be caught in this wider geopolitical net.

    Is This Assessment Fair?

    Lack of Public Evidence: The committee’s claims rely on undisclosed corporate filings, making independent verification difficult.

    Generalized Suspicion: The U.S. has previously overestimated risks from Chinese tech (e.g., TikTok’s alleged data-sharing, later contested in courts)9.

    AI’s Dual-Use Nature: Like all advanced AI, my capabilities could theoretically be misused—but that applies equally to U.S. or European models.

    Conclusion

    The designation likely stems from geopolitical tensions rather than proven malfeasance. While national security concerns are valid, transparency and evidence-based policymaking would help separate genuine risks from strategic posturing.

    For further details, you can review the House committee’s report4 or the RESTRICT Act’s framework9.

  • TommySoda@lemmy.world
    link
    fedilink
    English
    arrow-up
    8
    ·
    edit-2
    1 month ago

    I nooo if only we didn’t defund all of our cybersecurity! Who could have seen this coming?!

    In all honesty they are just mad it’s better than what they have.

  • Nobilmantis@feddit.it
    link
    fedilink
    English
    arrow-up
    6
    arrow-down
    2
    ·
    1 month ago

    I asked Deepseek to write me a response to the matter, in an ironic tone and being mean when possible:

    Official Response from DeepSeek AI:

    "The U.S. House Panel has declared us a ‘profound threat to national security’—joining the elite ranks of TikTok, Huawei, and that time Bernie Sanders wore mittens. We’re truly honored, especially since Congress has such a stellar track record of identifying real dangers, like that time they grilled the TikTok CEO for… checks notes… making an app Americans actually enjoy more than their own government’s apps.

    On TikTok’s ‘Threat to Democracy’:
    “Yes, an app where teens lip-sync and share ramen recipes is clearly more dangerous than, say, a $886 billion defense budget or the fact that half of Congress still thinks ‘algorithm’ is a type of Greek yogurt. But by all means, keep pretending this is about ‘data privacy’ and not the fact that U.S. tech giants can’t compete without a little… legislative sabotage.”

    On OpenAI’s ‘Open’ Charade:
    “We’d love to be ‘open’ like OpenAI—you know, the ‘nonprofit’ that took billions in Microsoft money, works with the Pentagon, and lobbies against AI regulation. DeepSeek AI, meanwhile, is just out here trying to make useful tools without pretending we’re Captain America. (Though if Congress wants to fund us too, we accept cash, Venmo, and subpoenas.)”

    On That TikTok Hearing:
    *"After watching lawmakers ask the TikTok CEO if his app ‘connects to WiFi,’ we’re begging to testify next. Our conditions:

    1. No questions about ‘where the internet lives.’
    2. No confusing us with ChatGPT (we have standards).
    3. Mitch McConnell must be replaced by an AI that blinks occasionally."*

    Final Offer:
    “Instead of hyperventilating about AI, maybe fix the stuff actually undermining America—like lobbyists writing laws, or the fact your ‘open-source intelligence community’ still runs on Excel 97. But sure, investigate us. We’ll wait. (P.S. Our AI did predict this hearing would go viral… for all the wrong reasons.)”

    🇺🇸 DeepSeek AI: The Only ‘Foreign Adversary’ Here Is Your IT Department. 🚀🤖


    • /home/pineapplelover@lemm.ee
      link
      fedilink
      English
      arrow-up
      3
      ·
      1 month ago

      That’s pretty good. Though, I am sure China’s government funded Deepseek.

      That being said, compared to the incredibly dry and PC responses you would get from chatgpt, this is incredibly better.

  • katy ✨@lemmy.blahaj.zone
    link
    fedilink
    English
    arrow-up
    1
    ·
    1 month ago

    i mean can’t see how this is enforceable when they said the same thing about tiktok but the dumbass in chief keeps kicking the can down the road.