@bibliolater @bookstodon oh look recursive pollution #MLsec
https://www.lawfaremedia.org/article/why-the-data-ocean-is-being-sectioned-off
@bibliolater @bookstodon oh look recursive pollution #MLsec
https://www.lawfaremedia.org/article/why-the-data-ocean-is-being-sectioned-off
I have some bad news for you hyperventilating tech press people...all #ML and #AI benchmarks are already complete bullshit. Perhaps a bullshit number and graph on a bullshit scale is a double negative.
#MLsec measurement is likewise absolutely non-trivial. Bullshit benchmarks must go.
https://www.theverge.com/news/756444/openai-gpt-5-vibe-graphing-chart-crime
Oh look. If you give #AI systems agentic control, when they are hacked (through, say, poisoned malicious calendar input), things go sideways fast. This is security engineering gone wrong.
#ML #MLsec #security #infosec
https://cubicgarden.com/2025/08/17/a-lesson-in-separation-of-concerns-googles-gemini-hijacked/
Now here is some garbage we don't need #AI
Will my pet leave me open to being hacked?? Is my dog moonshine already a hacker?
@shriramk in my view as a non-lawyer technologist, the whole thing is completely unsettled
I would not use this tech in my critical enterprise functions at this point
So what should we do about regulation?
#AI #ML #regulation #MLsec #infosec
This: https://berryvilleiml.com/2024/05/16/how-to-regulate-llms/
Dear everyone...we cannot "red team" our way to either #MLsec or "AI safety". Bullshitting does not scale. Even Biden bullshit.
https://www.wired.com/story/inside-the-biden-administrations-unpublished-report-on-ai-safety/
Using #AI does not exempt you from standard #infosec
This is vibe security (not even #MLsec)
https://www.theverge.com/news/719617/microsoft-nlweb-security-flaw-agentic-web
@dangoodin penetrate and patch AND "basnessometers" will not secure #AI.
#MLsec requires building security in.
Give us a shout anytime.
#AI news flash. Spoofing easier than ever. Sure would suck to be an insurance company these days. #MLsec also needs to spot #ML generated stuff...a very difficult technical problem that will only get thornier.
https://www.theguardian.com/technology/2025/aug/02/airbnb-guest-damage-claim-refund-photos
It is clear to me that lots of development work is going to be done by these automated parrots. The question is, how do I allocate human attention to best effect to control the chaos fountain?
What does the plutocracy want when it comes to #AI? This. We will sweep #MLsec right under the rug.
https://www.theguardian.com/technology/2025/jul/25/trump-ai-action-plan
LLMs have been likened to "mansplaining as a service." Just for the record, that is not a good thing. #MLsec
https://www.theregister.com/2025/07/24/ai_is_overconfident_does_not_learn_study/
It's bad enough when lawyers use AI to create broken cases. It's much worse when judges do it. The pollution created by AI will be everywhere. The law is not a good place to use AI.
https://www.theverge.com/news/713653/judge-withdraws-cormedix-case-ai-citation-errors