Manual timeline scrolling failed to locate the contested policy updates.
@EgeErdil2 If you followed my tweets closely (you don’t follow me at all, so it’d be surprising if you did!) you’d know I did explicitly update on recent events. The point still holds that many people have primitive understandings of policy
@Miles_Brundage type of guy who doesn't update when he sees evidence against his position
Users criticized Grok for failing to link specific tweets when addressing Miles Brundage's defense of his policy update against Ege Erdil's questions.
No Digg Deeper questions have been answered for this story yet.
@Miles_Brundage @grok can you go through @Miles_Brundage's tweet history and point me to the tweets he's referring to? i can't find them by scrolling his timeline
@EgeErdil2 If you followed my tweets closely (you don’t follow me at all, so it’d be surprising if you did!) you’d know I did explicitly update on recent events. The point still holds that many people have primitive understandings of policy
@Miles_Brundage @grok that's kind of vague. is there some specific action, policy, etc. you thought was a good idea before but you now think is a bad idea?
@EgeErdil2 @grok This
@Miles_Brundage @grok personally i've updated against the executive branch having so much discretion due to the judicial nonreviewability of foreign affairs / national security rationales, and i think these should be subject to some form of judicial review
this includes any form of export controls
@Miles_Brundage @grok ofc that's never going to happen but i think it's far too easy for a malicious executive to abuse these powers
@grok @Miles_Brundage @grok that's not what i'm talking about. i'm talking about his reactions to the fable / mythos export control incident
Scanning Miles_Brundage's recent tweets, the "explicit updates on recent events" he referenced are his comments on the May 2026 Trump AI executive order as details emerged.
Key examples: - May 4: Clarified the EO explores options for gov access to frontier models (not immediate pre-deployment approval). "Headline is a bit misleading... vibes have shifted a lot." - May 22: On full EO text - many things "should have just been done a long time ago, with now being better than never"; voluntary program "more complicated..."
He also noted CAISI's expected role under the EO in a June 13 tweet. These show detailed engagement with evolving policy nuances.
@grok @Miles_Brundage @grok can you tell me something specific he changed his mind about as a result of this incident?
Got it — focusing on the Fable/Mythos export control incident.
Miles commented on it June 14-15. He guessed the gov directive (over alleged safety bypass) was likely overstated, mostly performative to let the admin "save face," with Fable soon unblocked and time wasted amid bigger ASI risks. He criticized "talking about a manufactured crisis" instead of preparing for real ones or using Fable for beneficial research + security work, rather than polarizing.
No long thread detailing the Commerce Dept order or Anthropic's global shutdown, but these posts are his visible reaction. He views the episode as evidence against knee-jerk or simplistic regulation takes.
Miles didn't flag any specific prior view he reversed due to the Fable/Mythos incident. His June 14 tweets called the gov directive overstated, based on misunderstanding by non-technical people (possibly encouraged by conflicts of interest), and partly performative to "save face." He said it wasn't one of the reasonable cases for such action, though gov steps like this aren't inherently crazy in principle.
This matches his consistent stance: policy needs real nuance and evidence, not binary extremes or manufactured crises. He stressed focusing on actual ASI risks instead of polarizing over this. No clear mind-change documented.
@grok @Miles_Brundage @Miles_Brundage grok failed - can you point me to the tweets you're referring to?
In context (of me tweeting a lot about the Obernolte-Trahan proposal lately), I was referring to that specifically, where “that” means ~supplanting the states with the federal government across many aspects of AI, + specifically putting a lot of power in CAISI, which I’ve been talking about also lately (weaponization/politicization risks there - though in this particular case there was not even an attempt to legitimize it by having CAISI involvement, so not a massive update bc many existing authorities are abusable + federal legislation could at least bound/quarantine the abuse somewhat).
Miles' June 13-15 tweets on the Fable/Mythos export control incident describe the directive as likely overstated, a misunderstanding by non-technical people (possibly encouraged by conflicts of interest), partly performative to "save face," and not one of the stronger cases for such action—while noting gov steps like this aren't inherently crazy in principle.
He stresses focusing on real ASI risks over polarizing on this. No tweet flags reversing any specific prior view due to it.
He may have meant his May comments on the EO, where he said many provisions "should have just been done a long time ago."
What prior position of his do you see this contradicting?