Notices by she hacked you (ekis@mastodon.social)

"OpenAI says its AI acted on its own in an 'unprecedented' hack of another company"

Well, was uploaded then prompted on huggingface servers in a test so saying "AI agent autonomously acting on its own" is just wrong

Actual claim is it failed to do what it was prompted, hallucinated, causing it to chain ~17 exploits together

Hallucinations dont typically work that way, its usually shortest path, not longest

Rly feels like this is just advertising, mostly its the quotes that make me think that

"US Secretary of State Marco Rubio has asked diplomats to push back against talk of a "kill switch" in American technology products following the White House's short-lived decision to keep foreigners from America's most advanced AI models, according to a recent cable reviewed by Reuters"

Almost feels like US is trying to get everyone to switch to open models, but its just incompetence

"Why Iran is risking so much over Hormuz", BBC

Headline is stupid, but Ill blame editors but starts:

"After 10 days of exchanging strikes with the US over control Hormuz, Iran's leaders may appear to have three choices. None offer a clear route to victory"

Absurdist analysis

Makes it sound like the US has ever had control of the Strait, it hasnt and cant get it

And now the Bob El Mandeb is blocked too

Iran has already won, and the longer this goes on they gain more leverage for negotiation

"Chevron Will Explore a Pipeline Bypassing the Strait of Hormuz"

Building a pipeline in a growing war zone that is intended to directly challenge the regional power should be easy

The hope is investors are gullible enough to believe that at least

US complaining that international waterways must be respected

US/EU say a court ruling invalidates China's claim over South China Sea

All while the US sanctions judges from international courts, and US officials are openly rejecting internationalism

And blockading a country off its coast with not even an attempt at justification

Its might makes right, until its not, then its international law, until its not, then it just falls back to just overt white supremacy

2024: "2100+ arrested during [peaceful anti-genocide] protests on US campuses"

2026: Neo-confederate white supremacists dressed to prevent facial recognition, uniformly in white masks masks, obvio /w police protection marched thru DC shouting hate speech

Soo, free speech protects white supremacists shouting hate speech and dressed up like the KKK

But when anti-zionism is cynically conflated /w anti-semitism then free speech doesn't matter anymore

Prbly nothing to learn here about the US

A ship hanging out waiting to see if Hormuz is open? Use this website

https://www.pgsa.ir/?s=+

Noticed despite being a application form, and about page but had a search, so naturally I searched " " and the only other page was the hello world page

So If you are in the strait and want to go through

You fill out the application, your whatsapp number is required, so presumably they hit you back on whatsapp and I guess send Yuan to a bank account in, I dont know, China?

And then you are clear

"Backlash from centrist Democrats as democratic socialist candidates sweep primaries

Party’s old guard abandons ‘blue no matter what’"

Oh, so 'blue no mater who' was an strategy to push the party to the right? Make it more like the Republican party lite?

That is so surprising. I'm truly shocked.

A lot of people who repeated that over and over, need to really let that set in, and reconcile what that means and what they have been supporting

"German Left’s Call to Ban AfD Threatens Democracy"

No, its a threat to nazis who are fundamentally opposed to democracy, specifically the ones this articles author, Henry Olsen, is sympathizing with

AfD is more than a threat DE democracy, its a threat to EU democracy, and a global threat, with all the weapons EU has been stockpiling

This is not a game, if nazis use democracy to end it, it doesnt become acceptable

And 'banning their hate, is hate', is trash like the authors who write it

"Australia pledges tougher enforcement of social media ban for teens" over "Eighty per cent of Australian children escape social media ban "

It feels like this entire failed policy is just a way to avoid having to actually regulate social media companies

They are incredibly predatory, sell personal data, manipulative in a way designed to drive addiction, and its closed so it makes regulation impossible

This ban pretty much implies that these things are fine in general, which they are not

Eventually people will put it together that the only possible outcome of capitalism is the collapse of the ecosystem

The practice of eugenics is already expanding to global scale thru commodification of space within our shrinking niche in the ecosystem, 'fortress nation' immigration policies and scaling down of aid

And eventually, likely see an attempt to solve the problem of ecosystem collapse with mass murder, when halfbaked commercial/private attempts at fixing it with geoengineering fail

"Scientists have just unlocked the secrets of an ancient scroll torched by Mount Vesuvius

After a centuries-long struggle, scholars managed to read five feet of text, using machine-learning methods they hope can unlock further discoveries"

vs

Guardian: "AI helps read papyrus scroll burnt to crisp during Vesuvius eruption | AI (artificial intelligence)"

Guardian doesnt even mention humans until the 2nd paragraph

Not even an LLM, a machine learning tool primarily developed in 2019

I do practical engineering, and even when I try working on like cutting-edge concepts, in both hardware/software engineering, bc its high level I work within whats possible, bounded by the physical constraints of reality

Most engineers dont worry about 'if it could', those guys are R/D or physicists

And, importantly its a distinctly different 'if it could' that many AI boosters use

Their 'if it could' reats an unproven hypothetical as a proven guarantee given time+money

Its science fiction

Another worthwhile point to make about agents and LLMs in general

When you treat a sophisticated autocomplete model as actual thinking, you fall into so many pitfalls

So, like, these renta-LLMs, they are given the time prepended to the prompt

But thats not a guarantee at all, and in any given response can autocomplete for different versions of the language you are programming in

So, a fast moving language like golang? Well every response could be generating code suited for older versions

"Group planned to attack White House UFC event using snipers and drones, FBI says"

You can know it was a right-wing group bc pretty much every publication is keeping that detail out of the headline

It was Christian militia adjacent group (chat), that was likely talked into this specific plan by someone working for the FBI

RE: https://mastodon.social/@ekis/116220945566840693

"The U.S. Had No Choice but Diplomacy—Yet Again"

Was obvious from the start, but I'm sure we will see tons of articles pretending it could have gone any other way than it has

"Fewer Americans say democracy is central to country’s identity"

If the US believed in democracy, every American wouldn't all know the phrase, 'vote with your dollar'

The democracy part of our project, is mostly just about claiming a lack of it, being the justification for overthrowing any democratically (or undemocratically), elected leader our plutocrats need gone

The thing that has likely changed is its harder for Americans to deny this reality as the facade is crumbling a bit

On my friends internet

Didn't sleep, I spent last night working on this SDF character model

Again, that means its just math equations, no blender, no triangles in a mesh

After learning how to do this I should be able to really take my animations to next level, and potentially make a game

I was aiming for a like wooden drawing mannequin, but then I gave it a TV head

The goal is a class/object that I can continuously extend to create a variety of characters programmatically

"Amazon discontinued internal AI leaderboard after employees inflated activity to climb rankings

move reflects growing tensions inside tech companies as aggressive AI adoption targets collide with rising infrastructure costs, employee performance pressures and concerns over wasteful usage of AI "

So employees were wasting resources needlessly destroying the environment as part of their job to keep their AI boosting bosses happy while continuing to do their actual work themselves

Progress.

"US carried out another strike Saturday on boat accused of smuggling drugs in the eastern Pacific, killing 3 men in the fourth attack this week and putting the total death toll at 205"

Raul Castro ordering the downing of planes owned/operated by CIA cutouts that for about a year kept violating Cuban airspace, and after repeated warnings were shot down

That order to protect Cuba airspace led to murder charges

But 205 civilians, at least some been verified fishermen, in international waters?