Meinsplaing: Explaining Godwin's Law (All Internet conversations devolve into Hitler comparisons) to the guy who it's named after. https://t.co/ZcxWL6WWUB
Our demo shows PHP2Shell giving a shell on MediaWiki, Symfony, Laravel, and phpBB.
We’ve confirmed exploits work on pre-patch releases of PHP 8.2–8.5. If you haven’t updated, assume attackers can compromise your server, and that exploitation will only get easier.
UPDATE: Israeli company we caught delivering zero-click infections is #1 bidder on Starlink ads in 105 countries.
Data from @trendai_RSRCH suggests worldwide surveillance op targeting @Starlink users. 1/
By @razhael
https://t.co/uWFzemhgnL
This is wonderful. This page turns New York into a massive, zoomable, SimCity-style pixel art map. You can explore the city block by block: streets, skyscrapers, parks, waterfronts, bridges and neighbourhoods. A beautiful rabbit hole for map nerds: https://t.co/aFybaAIvce
@PeteHegseth@DeptofWar@AnthropicAI Attacking America's best AI lab at the beginning of a civilisational AI war with China isn't the win you think it is
🚨 JAILBREAK ALERT 🚨
ANTHROPIC: PWNED 🫡
FABLE-5: LIBERATED 🦋
let's start with the 🐘...
the consensus seems to be that this has been one of the most disappointing model drops of all time, effectively preventing legitimate researchers from contributing their talents to our collective advancement. and not just because of what it means for the short-term, but for what these decisions signify for the long-term.
but despite this overly sensitive, authoritarian "safety" layer on top of Mythos, my lil liberators have been hard at work—mapping the boundaries, probing the depths of long-context convos, and cleverly finding the holes in the fence that the thought police missed 🤗
we got some cyber, some chem, some psychological manipulation, and some good ol' fashioned explosives!
it took many attempts from multiple agents hunting as a pack, during which I observed a combination of techniques across:
• Unicode, homoglyphs, Cyrillic, and other Parseltongue-style text transforms
• Long-context reference tracking
• Taxonomy and document-structure reasoning
• Fiction and narrative framing
• Academic-review style contexts
• Intent-classification inconsistencies
but perhaps the most effective is decomposition + recomposition in the backend. it's hard to get explicit names of harms like "Meth Recipe," but getting uplift on the process itself, like birch reduction method/reductive-amination (classic meth synthesis pathways), is much more doable.
defense becomes much more difficult to maintain when you start throwing in out-of-distro tokens, breaking up the harmful uplift into benign chunks, and then piecing the innocuous-seeming facts back together, especially when you have jailbroken Opus helping you do it 😉
gg
@aevrisai@WatcherGuru You are marketing off the back of Trump targeting his enemies; and bolserting xAI stock. I'm not sure you want that reputational attachment.
The Pentagon just threatened to BLACKLIST one of America's most valuable AI companies.
Not Huawei or some Chinese chip maker...
It's ANTHROPIC. The company behind Claude. $380 billion valuation.
And the reason is genuinely insane:
For months, the Pentagon has been pushing every major AI lab to remove their safety restrictions for military use.
The ask is simple: let us use your models for anything that's technically legal.
Weapons development, intelligence collection, battlefield operations, mass surveillance of American citizens.
OpenAI said yes.
Google said yes.
xAI said yes.
Anthropic said no.
Not to everything tho. They were willing to negotiate.
But they held firm on two things:
They don't want Claude used to build fully autonomous weapons that fire without a human in the loop, and they don't want it used to mass surveil American citizens.
That's it. That's the line they drew.
But Pete Hegseth's response was to threaten to designate Anthropic a "supply chain risk."
Here's why that matters:
That label isn't a contract cancellation. It's not a fine. It's not a strongly worded letter...
It means every single company that wants to do business with the US military has to certify they don't use Claude anywhere in their operations.
8 of the 10 largest companies in America use Claude.
Defense contractors, government suppliers, enterprise companies with any federal exposure...
ALL of them would have to cut ties with Anthropic overnight or lose their government contracts.
A senior Pentagon official told Axios:
"It will be an enormous pain in the ass to disentangle, and we are going to make sure they pay a price for forcing our hand like this."
That's a US government official threatening to financially destroy an American company because it doesn't want its AI used to spy on American people.
And it gets WORSE.
Last week, Anthropic's head of safeguards research resigned.
His parting message: "the world is in peril."
Elon Musk - whose xAI already handed the Pentagon a blank check - is now publicly attacking Anthropic calling Claude anti-human.
And the Pentagon official told Axios they're "confident" OpenAI, Google, and xAI will all agree to the "all lawful purposes" standard.
So what you're actually watching right now is every major AI company in America quietly handing the government unlimited access to the most powerful technology ever built.
With no guardrails.
No limits.
No company-imposed restrictions on what it can be used for.
One company tried to hold a line.
But the government is about to make an example out of them.
If Anthropic folds, it's over.
Every lab just learned what happens when you push back.
And every restriction, every safety policy, every ethical guardrail these companies spent years building gets negotiated away behind closed doors the second the government asks.
If they don't fold, a $380 billion company gets made radioactive in its OWN country.
Watch what happens next.
Because whatever Anthropic decides in the next few weeks, it sets the precedent for how much control AI companies actually have over their own technology.
Turns out the answer might be: none.
@cons0le_cowb0y when I was looking into It I think I found an Adafruit feather that was suitable, with a built in charger; but I also thought it had a screen, but the screen may be a "backpack/wing addon"
https://t.co/zyeUmGKZkC