This seems like a weird way for Marc Andreessen to acknowledge that supporting Trump was a mistake but I'll take it!
8:36 AM · Jun 26, 2026 · 21.5K Views
Zvi Mowshowitz argues the new White House AI policy allows ad hoc control over frontier AI model development
Story Overview
Zvi Mowshowitz calls out the June 2026 executive order for creating a classified benchmarking system that lets agencies pick which frontier models count as covered and then negotiate early government access on their own timeline, turning what was meant to be light-touch security support into opaque scheduling power over releases like GPT-5.6.
19627282325.8K
Policy Risk
Deregulation voices now tied to rollout disorder
Journalists trace the ad-hoc pattern directly to earlier pushes against formal rules, leaving agencies without templates and forcing one-off calls that stretch timelines for any model tagged with advanced cyber risks.
Open Question
Model teams left guessing at approval windows
The voluntary framework gives no fixed calendar, so labs must wait on agency determinations that can shift per release and per partner list, with nothing in the order spelling out how many days or which criteria decide the stagger.
Sentiment
Users criticize Andreessen's Trump support and deregulatory AI stance for enabling policy chaos from a self-absorbed figure while sarcastically questioning his introspection or motives, though some agree he shows indirect regret.
Pos
16.7%
Neg
83.3%
10 comments with sentiment.
Cluster Engagement
Digg Deeper
No Digg Deeper questions have been answered for this story yet.
Posts from X
Most Activity
Most Activity
VIEWS2.2KLIKES42RETWEETS3REPLIES3
Timothy B. Lee@binarybits
People are saying "no he's criticizing Dario for scaring Trump about frontier models." But whatever you want to say about Dario, the person most responsible for Donald Trump's policies is Donald Trump. And people who helped get Donald Trump elected also bear some responsibility.
Timothy B. Lee@binarybits
This seems like a weird way for Marc Andreessen to acknowledge that supporting Trump was a mistake but I'll take it!
27mViews 2.2KLikes 42Bookmarks 0
Timothy B. Lee@binarybits
@DkSemyon I'm sure that's what he means but that's ridiculous. The entity that did this is the Donald Trump Administration, which Andreessen supported and most of the EAs and AI safety people did not.
1hViews 458Likes 22
Dima Kill-Steal@DkSemyon
@binarybits Nah, I think he's probably blaming it on EAs, Frontier labs, woke, AI safety ppl, etc. They're the woman asking the question in the meme.
1hViews 538Likes 19
Andrew Mayne@AndrewMayne
@binarybits This is a reference to Dario hyping Mythos fear.
Timothy B. Lee@binarybits
This seems like a weird way for Marc Andreessen to acknowledge that supporting Trump was a mistake but I'll take it!
1hViews 650Likes 11Bookmarks 0
Timothy B. Lee@binarybits
@AndrewMayne Yes I know. But he didn't do it!
54mViews 132Likes 3
Shakeel@ShakeelHashim
@deanwball put it very well this morning. https://www.hyperdimensional.co/p/what-should-be-done
3hViews 114Likes 6
Timothy B. Lee@binarybits
@PersonaIData No.
1hViews 484Likes 3
Shakeel@ShakeelHashim
And I wrote something similar a couple weeks ago: https://www.transformernews.ai/p/trumps-ai-executive-order-was-inevitable
3hViews 200Likes 7
Kingston Jr@PersonaIData
@binarybits Did he delete this?
1hViews 527
Timothy B. Lee@binarybits
@subniferus @DkSemyon Donald Trump and his subordinates are grownups who are responsible for their own decisions.
40mViews 23Likes 2
Alexander McCoy@AlexanderMcCoy4
@ShakeelHashim Wait is that what he means????
I (naively) assumed his post was a fleeting moment of self-awareness!
1hViews 71
lovable rogue@lovabler0gue
@binarybits 1. this is directly caused by dario scaremongering and refusing to play ball with the army 2. the dems explicitly said they’d regulate AI even more aggressively than trump is doing right now
7mViews 11
Yags@yaghsizian
@binarybits If only he was capable of introspection…
1hViews 91Likes 2
Aeneas@IM_Aeneas
@binarybits @pmarca response?
1hViews 225Likes 1
Arthur Rock@iheartsophon
AI safety people: Frontier AIs are weapons and should be kept out of the hands of bad actors
US government: ok well that’s our jurisdiction then, we’ll decide who can have access
AI saftiests; no not like that! We meant we need the government-like power to decide who has access even though we’re unelected!
1hViews 119Likes 1
Kingston Jr@PersonaIData
@binarybits Weird couldn’t find it
1hViews 95Likes 1
Timothy B. Lee@binarybits
@AndrewMayne Like does Marc Andreessen think Donald Trump is a child?
Timothy B. Lee@binarybits
@AndrewMayne Yes I know. But he didn't do it!
53mViews 41Likes 1Bookmarks 0
Andrew Mayne@AndrewMayne
@binarybits ?
49mViews 39Likes 1
Aashish Reddy@_AashishReddy
@ShakeelHashim
2hViews 24Likes 1
Cajun Bobby@brsgr4049
@binarybits @DkSemyon There is an invincible, non-falsifiable cope that Harris would have basically banned AI outright that these types can retreat to
They may pivot next presidential cycle, but they'll back the admin for the foreseeable future
58mViews 14Likes 1
Load more posts