in the age of artificial intelligence, non-digital archives to become priceless

[gadget42]

in the age of artificial intelligence, non-digital archives to become priceless

https://jukkan.com/openclaw-open-the-front-door/

https://www.whodoyouthinkyouaremagazine.com/news/gemini-artificial-intelligence-the-national-archives-fake-records

also this might be interesting as well:

https://www.scientificamerican.com/article/ai-slop-how-every-media-revolution-breeds-rubbish-and-art/

[mocore]

thanks for the links

interesting reading
for our "interesting" times !

[mocore]

:|

>Claude code source "leaks" in a mapfile
https://neuromatch.social/@jonny/116328988554490524


via : https://forums.freebsd.org/threads/claude-code-cracks-freebsd-within-four-hours.102251/

[nick65go]

via : https://forums.freebsd.org/threads/claude-code-cracks-freebsd-within-four-hours.102251/

someone said that " Claude Code cracks FreeBSD within four hours? No it didn't. It wrote code to exploit a known CVE given to it within four hours."
Whatever, but it worries me that a smaller code base (FreeBSD) than linux code base is (or will be) vulnerable at a small cost for any bad actor (not only for state agencies) given the fast progress of A.I.

https://www.forbes.com/sites/amirhusain/2026/04/01/ai-just-hacked-one-of-the-worlds-most-secure-operating-systems/
"Regulatory frameworks will need to adapt. Current compliance regimes assume that security is measurable through checklists and periodic audits. They do not account for adversaries whose capabilities compound with each improvement in foundation models. Security regulations written for a world of human-speed attacks are inadequate for AI-speed threats.Fully autonomous cycles of bug identification, fuzzing, exploit generation, exploit application and exfiltration or destruction can drive a new lightning war."

[mocore]

before
>Current compliance regimes assume that security is measurable through checklists and periodic audits.

now they *should* assume  xyz
about a term coined in a grant application to the rockefellerfoundation  :0

idk but...
perhaps the the shape of the *real* problem is out lined by common theme ?

|>By definition, all assumptions introduce possibilities for error;
| - https://en.wikipedia.org/wiki/Occam%27s_razor

then
>Security regulations written for a world of human-speed attacks
like manual typing typing each port number ?
and pressing enter ??

and now
"ai"/open claw  can write a for loop *for you* 

tbh im struggling to make seance of it all  ( though fear not : in my case this is "normal" )


>given the fast progress of A.I
last time i checked** ( ftr im just some idiot ( member of a self destructive species[0] )  with a keyboard mind you)

ppl writing ML ( note not the marketing term but academic discipline ; aka *apparently* not perused solely for the purpose of financial gain  )  papers
seamed to have some doubts about the speed of "progress" in this regard

** ( Q: Why Do Keynote Speakers Keep Suggesting That Improving Security Is Possible?
A: Because Keynote Speakers Make Bad Life Decisions and Are Poor Role Models
 https://www.usenix.org/conference/usenixsecurity18/presentation/mickens )
 
[ 0 ] - https://en.wikipedia.org/wiki/Tetraethyllead
( decades of putting this into *our* environment could have *some* beneficial outcomes
,... if and only if  you value its apparent impulsive / aggressive /etc.. effects on cognitive development    )

Whatever, but it worries me

if you bring the fear
ill bring the uncertainty and doubt

at least closed source alternatives wont get this kind of public scrutiny 

so lets assume all that obscurity is a good thing!
..install party *anyone*  ?

[press return]

[nick65go]

ill bring the uncertainty and doubt
at least closed source alternatives wont get this kind of public scrutiny 
so lets assume all that obscurity is a good thing!

Errare humanum est ... As stoics said, someone should not be afraid for something out of his control (like software development made by others for not so skilled ones). Of course close source software is a mistake, (security by obscurity fail) derived from intellectual property rights on "common" knowledge (cowboy style: first come takes all).

My assumption was that smaller software has less "attack" surface, therefore more secure when carefully crafted. Unfortunately (some, most?) developers adopted the tactic of releasing fast + unpolished (for testing).

AI is about mostly automation, which could have happening long time before. But with so much money invested in hardware so now is faster to brute force many paths. And find mistakes. Which is good to know, that gods are not invincible just because they are big and loaded.

[mocore]

My assumption was that smaller software has less "attack" surface, therefore more secure when carefully crafted.

i have to admit i found my way to tcl making similar assumptions

in my case it was less about "attack" surface and more about

likelihood i could make reasonably accurate assumptions  about "system function"
... in the rather idealist hope of avoiding **`function failure` ( though activeX(ploit) did play a large part in that fwiw )

 ** which was the main reason i have any cause to involve my self in this world of digital torment     

 perusing the minimal functionality principle in to the domain of programing langues
 lead me to lisp , and associated academic CS en-devours
 
 & while pulling theoretic strings of set theory  and abit of  FP
 finally landed on unfortunate conclusions of kurt's https://en.wikipedia.org/wiki/G%C3%B6del%27s_completeness_theorem

*irrelevant tangent*
https://www.youtube.com/watch?v=THCj2AJuNVE (4 Hero - Mr Kirk's Nightmare)

and found out what happens when you combine
cats and curiosity
 :'(

all of this oddly makes me recall a quote from garry oldman in leon
"I haven't got time for this Mickey Mouse bullshit!"
 

amongst *all this* increasing maddening complexity
 the core philosophy still shines though
 from the perspective of user script-ability / modify-ability  !