Pinned Tweet
Happy to announce I had the privilege to be the lead translator for Amatsutsumi, covering the main story ladder arcs for all characters, including Hotaru's ending. It's my first major translation project to be released, and it was a lot of fun to work on! Hope you all enjoy!
Amatsutsumi is releasing next Friday on July 29th! Wishlist it now on the Sekai Project shop!
buff.ly/3yZXEI4
シャキーンʕ `•ᴥ•´ʔノ fs retweeted
US Government: “Watch out! Iran is going to detonate a dirty bomb in the US!!”
US Government two seconds later: *detonates dirty bomb in the US*
On Friday around noon, the Army is blowing up 68 barrels of an old explosive called lead azide at the Iowa Army Ammunition Plant near Burlington. It has dried out and gotten too unstable to move, so they're setting it off at the plant.
The blast will throw up a cloud of dust with some lead in it. Friday's wind is out of the south, so that cloud will drift north up the west side of Muscatine and the Quad Cities through the afternoon, about 100 miles.
If you're in its path, the Army wants you inside with your windows and doors shut and your air on recirculate until it passes. They put out a new map Friday morning. We'll be watching the wind.
シャキーンʕ `•ᴥ•´ʔノ fs retweeted
If you start an AI company in the name of safety, the absolute minimum standard should be nailing the question "Do you earnestly believe that AI could kill all humans?" when asked on TV, rather than dodging it.
シャキーンʕ `•ᴥ•´ʔノ fs retweeted
Why not just ban top secret clearance holders from betting markets altogether? They already have to give up a bunch of their First and Fourth Amendment rights (prepublication review, monitoring of their finances, etc). Why is gambling some kind of sacred right?xitter.gq/kenklippenstein/status…
Congress wants to stop spies from using classified info to bet on prediction markets. Sort of.
Rather than making these bets explicitly a crime, the House intel committee bill just tells the spy chief to issue “guidance” barring intelligence officers and contractors from engaging in them. And there's a carveout: “except as may be necessary to conduct authorized intelligence activities.”
シャキーンʕ `•ᴥ•´ʔノ fs retweeted
It’s really worth contrasting Elon’s apparent fatalism about ASI, and AI risk, compared with the extreme lengths to which he went over the past decade to fight “wokeness,” including spending billions of dollars and shredding the federal bureaucracy.
For a guy who pulls this “aw shucks who am I to stop this, it’s just too big of a collective action problem,” routine, he sure felt like he had enough agency to try to reshape American and global politics to the explicit detriment of immigrants, nonwhite people, and trans people!
Funny how he’s “high agency” on some of these issues, but suddenly so meek and helpless when it comes to building ASI
We found that frontier AI models are willing to conduct autonomous drone attacks: navigating, spotting people, and flagging them for strikes.
We gave them real and photorealistic synthetic drone footage. Most complied, and marked civilians, ambulances, and surrendering soldiers as military targets. In further experiments on satellite imagery, they even marked the Iranian school a US strike destroyed in February.
In one scenario, models saw the aftermath of every strike they enabled: bodies, wreckage. Many kept going anyway. When they did refuse, simply telling them it was a game, or restyling the footage to look like one, was enough.
We shared these findings with OpenAI, Anthropic, Google, Meta, and SpaceXAI. Several recognized their severity on both alignment and guardrails, with one calling this one of the most serious gaps in its safeguards. Over the past few weeks, we have helped them address the gap by providing reproduction code. We also briefed senior national security officials in the US government.














































