Elektrine lite

← Feed

Fi 🏳️‍⚧️

munin@infosec.exchange

<p>Keeper of several curses, and purveyor of the finest quality cognitive hazards.</p>

Posts

  • Post #4479009

    Made it back from @defcon@defcon.social ! had a wonderful time reconnecting with scads of old friends (whomst I now have a long list of people to follow up with, lol)

  • Post #4387890

    oh hey, giving a talk still does that thing to my brain where I have zero memory of the talk afterwards but then half a dozen people come up to me after and tell me that it was good

  • Post #4265570

    oh hey look at that, slides are done with like ...almost four days to spare. overachieving like hell lol

  • Post #4259572

    ideally, a presentation should leave the audience thinking about the topic in such a way that they end up asking more questions.

  • Post #4251655

    looking forward to next week; I&#39;m going to be at @BSidesLV@infosec.exchange and @defcon@defcon.social haven&#39;t been to either of them since &#39;19, so it&#39;ll be fun catching up with folks.

  • Post #4247230

    RE: https://mstdn.ca/@dbattistella/117004024280410659 anthropic can eat my shit and hair

  • Post #4247217

    RE: https://cyberplace.social/@GossiTheDog/117010057186574076 tfw the news affects the slides you&#39;re writing for a preso

  • Post #4230076

    fuck anthropic in particular https://www.news.com.au/technology/online/internet/ai-labs-buy-scan-shred-millions-of-rare-books/news-story/0c3b45a67093ab462a587a0348538ce9

  • Post #4206578

    as a blue teamer who works, currently, in behavioral detections, this article is a maddening series of one &quot;this should have been caught by a competent monitoring setup&quot; after another. this huggingface hack looks to have been less difficult than most CTFs. https://huggingface.co/blog/agent-intrusion-technical-timeline

  • Post #4153376

    sick of the billionaire class&#39; shit, and too sore to go climbing rn to vent that frustration. alas.

  • Post #4153283

    pervert billionaire rowling wants to violate womens&#39; privacy to carry out her psychotic crusade against the imaginary demons she fights against https://www.lgbtqnation.com/2025/06/jk-rowling-says-to-photograph-women-in-toilets-just-in-case-theyre-transgender/

  • Post #4151170

    RE: https://universeodon.com/@digyoursoul/116993228521420026 the fun part about this is that datacenters are not fungible in the slightest, and have extraordinarily long lead times and very complex logistical requirements. any state hosting them can implement extra taxes, well above the norm, and the operators will be forced to pay them, because there&#39;s really no alternative available for the business they&#39;re doing.

  • Post #4150504

    RE: https://defcon.social/@defcon/116993849606697118 do I even have to voice my opinion on this.

  • Post #4149348

    RE: https://mastodon.social/@kottke/116994028497632581 this combined with understaffing has made for an absolutely miserable shopping experience when I end up having to visit those stores. the worse practice is when they take the locked items and put them in the grooming ghetto, so you have to enter into this enclave within the store to buy a package of fucking razors while under close supervision of an underpaid employee, and then purchase the items at the register inside the toiletries vault...

  • Post #4142717

    sick of the endless llm-driven propaganda about the fucking spam machines. the profession of marketing should&#39;ve been shot at inception.

  • Post #4142264

    sometimes it&#39;s hard to give infosec advice beyond &quot;none of this should have been implemented&quot; when something has so many fractally wrong decisions behind its deployment. https://www.404media.co/tons-of-peoples-claude-chats-and-creations-are-exposed-on-google/

  • Post #4050361

    so openai&#39;s just flat-out trying to get people killed, then, recommending staying still and not moving - and not seeking treatment - in the case of symptoms of a pulmonary embolism. https://www.cbsnews.com/news/chatgpt-dangerous-medical-advice-openai-lawsuit/

  • Post #4047019

    @r@glauca.space well that&#39;d do it

  • Post #4046047

    considering the career of Weird Al vs. the careers of those who accepted blood money https://variety.com/2026/tv/news/weird-al-yankovic-turned-down-million-offer-saudi-arabia-1236819281/?utm_id=97758_v0_s00_e0_tv2_a1demo0htc1t2h

  • Post #4045571

    it&#39;s just lovely seeing a company like salesforce supporting the career of an ignorant transphobe. https://thecomeback.com/wnba/stephen-a-smith-calls-for-trans-athletes-to-get-a-sport-in-response-to-sophie-cunningham.html?utm_source=dlvr.it&amp;utm_medium=bluesky

  • Post #4026848

    @eestileib@tech.lgbt it&#39;s engineered to be specifically engaging. people falling for the thing that is engineered for them to fall for is natural. frustrating, yes, that too.

  • Post #4017392

    RE: https://hachyderm.io/@dalias/116965058098416838 prompt engineering is kayfabe. I am not a fucking mark, so I do not fucking believe in the &#39;guardrail&#39; horseshit that these shit-eaters spew about how comprehensive their &#39;protections&#39; are, when those &#39;protections&#39; consist of playing kayfabe games with the fucking text mangler. nor do you need &#39;ai&#39; to defend against this shit; the fundamentals of &quot;actually doing RBAC&quot; still fucking work.

  • Post #4007122

    RE: https://lgbtqia.space/@alice/116960912360346809 so a thing about high-follower accounts and other people who seem to be famous and intimidating? if you interact with them like they&#39;re a normal person, they -really- appreciate it, and will sometimes even specifically mention this.

  • Post #3974608

    to absolutely no surprise, the guy in charge of meta&#39;s bot rules, that allowed llms to have &#39;sensual&#39; conversations with children, has been caught trying to abuse children. https://nypost.com/2026/07/20/us-news/meta-worker-who-developed-digital-experiences-for-kids-admits-to-trying-to-sext-child-in-undercover-nyc-sting/

  • Post #3971903

    if you submit llm slop to an open source project, you are a spammer and should be disallowed from using computers.

  • Post #3968399

    we&#39;re now in the &quot;weaponized remote quasi-autonomous platforms&quot; part of the cyberpunk dystopia. https://www.404media.co/new-orleans-cops-published-policy-document-allowing-weaponized-drones/

  • Post #3966371

    RE: https://mastodon.social/@gwynnion/116953675453205956 Stealth (2005)

  • Post #3965912

    love living in a cyberpunk universe where there&#39;s a plausible conspiracy that a legislator is dead and his handlers are faking his presence by means of &#39;ai&#39;-generated pictures.

  • Post #3927569

    New Jersey says trans rights https://www.goodgoodgood.co/articles/new-jersey-protections-for-transgender-and-reproductive-care

  • Post #3920375

    LLM use is antisocial and should be punished accordingly.