Brian McGrail

AI policy/law. Attorney. Former @CommerceGov.

Views are mine. But you can have them if you want?

[Daniel Kokotajlo @DKokotajlo]

In AI 2027, we predicted that AI would take over the world or irreversibly concentrate power.

In AI 2040: Plan A, we've laid out our positive vision for what should happen instead.

[Brian McGrail @brianmcgrail]

Charlie's piece is predictably great but naturally I take his discussion of preemption as vindication of *my* take that the only thing we know for sure about this provision is that we'll have tons of litigation if this passes

[Brian McGrail @brianmcgrail]

The one thing that can be said with confidence about the scope of the preemption provision in Obernolte/Trahan is that it's very unclear - so unclear, in fact, that it would take years of litigation before we know what is preempted if this passes.

To start, it is simply impossible to draw a clean line between "development" and "deployment" for modern AI systems.

Then consider that industry will have the best lawyers money can buy cooking up arguments (quite possibly including args no one has even thought of yet) that the preemption provision sweeps as broadly as possible.

Keep in mind, too, that this provision will be interpreted by judges who - I regret to inform you - are not exactly on the bleeding edge in terms of understanding AI.

Put it all together and it's anyone's guess how this provision will be interpreted.

I'm sure there will be (likely already are) bill supporters claiming that AI safety advocates like me are mischaracterizing the scope of the preemption provision. So just to be crystal clear: I'm not confident how this would be interpreted. In fact, my main point is that I think you should be skeptical of anyone claiming that they are confident about this.

[Brian McGrail @brianmcgrail]

Can someone explain why you are allowed to have a baby on your lap but can't have a laptop open during takeoff and landing?

[Brian McGrail @brianmcgrail]

Huawei is nanoseconds behind!

[Financial Times @FT]

Nvidia’s banned AI chips double in price on China’s black market https://t.co/mg2yQSQENk

[Brian McGrail @brianmcgrail]

If you thought the populist AI backlash was big before...

[unusual_whales @unusual_whales]

JUST IN: AOC says Congress should break up Apple, $AAPL, amid looming price hikes for MacBooks and iPhones

[The Tech Oversight Project @Tech_Oversight]

NEW: Today, TOP launched a first-of-its-kind AI Super PAC research book and ad library to provide journalists, researchers, and the public with real-time campaign finance data, political research, and a plethora of ads Big Tech is running across the country.

🧵 More details ⬇️

[Nathan Calvin @_NathanCalvin]

NY-12 retrospective in the NYT from Teddy Schleifer.

Hard to imagine this was the headline that Leading the Future wanted when they went after Alex Bores with overwhelming force in a crowded field six months ago.

[Teddy Schleifer @teddyschleifer]

Alex Bores on the AI industry fighting him:

"I didn't get in this race to make a point about AI. But some of the most powerful people on the planet — a handful of oligarchs hellbent on preventing any regulation of their industry whatsoever, any check on their power, the very people who are fueling Donald Trump — decided they wanted to make me an example in this race."

"This was a huge and unprecedented fight, and we did not back down."

"Future victories will be built on the shoulders of the progress of this campaign. That's how movements work."

[Will Poff-Webster @willpoffwebster]

This is the first national housing supply package to pass in my lifetime. Members and staff did an incredible job getting to yes on a combined bill includes more than either chamber passed on its own.

Among ROAD to Housing's over 40 provisions:

1. It removes the onerous steel chassis requirement from manufactured housing.

2. It preserves new build-to-rent housing supply for renters.

3. It exempts infill housing from environmental review.

4. It penalizes expensive communities that don't build and rewards communities that allow more homes, putting federal spending on the side of pro-housing policy.

5. It provides technical assistance for communities seeking to reform their zoning and land use regulations.

6. It authorizes resources for housing preservation, funding repairs for low-income homeowners and small landlords.

7. It includes rental assistance, disaster recovery, and public housing reforms.

There are too many good things in here to count, and too many people to thank.

[Jay Shooster @JayShooster]

At least one of the candidates with the strongest AI platforms in the country has already won tonight!

Congrats to @bradlander!

[Brian McGrail @brianmcgrail]

crazymaking that the same admin that is objecting here also championed selling chips to G42 in the UAE, which has well documented ties to China

[WIRED @WIRED]

SCOOP: A South Korean telecom giant’s alleged links to China were reportedly at the root of tensions with the White House over export controls on Anthropic’s most powerful AI technology. https://t.co/FW42JkBhBR

[Nathan Calvin @_NathanCalvin]

Dean Ball going to OpenAI to lead "Strategic Futures," a new team reporting directly to CSO Jason Kwon, with a role "shaping frontier AI policy" including "proposals for legislation."

Seems quite notable that the role will not be reporting to Global Affairs head Chris Lehane, but instead directly to Jason Kwon. I would have previously expected shaping frontier policy legislation to fall primarily within Global Affairs (and therefore Lehane's) purview.

Only time will tell, but my initial reaction is that I hope and cautiously expect that Dean's new role will lead to more of things like OpenAI supporting SB 315 and advocating for thoughtful policy on auditing, gene synthesis screening, and support for CAISI, and less of the OpenAI that advocates for complete liability shields or underplays AI capabilities and risks when talking to policymakers.

It's also great to see top technical researchers like Noam Brown taking interest in OpenAI's frontier AI policy work. The future of AI policy is extremely important and high stakes, and we need technical folks to be deeply engaged in policy conversations for this to go well.

[Brian McGrail @brianmcgrail]

@DavidSacks Why raise Anthropic's "track record" in your posts about Mythos/Fable if you thought it was so important for the world to take the underlying threat seriously?

[Chris McGuire @ChrisRMcGuire]

I share concerns about China’s access to advanced AI models, but if the admin feels so strongly about this, I have a series of questions it should answer:

- Why did it loosen export controls to allow AI chip sales to China, which allow China to build its own Mythos?
- Why is it not enforcing existing export controls that would prevent China from smuggling AI chips from Southeast Asia and other countries?
- Why is it not enforcing existing export controls that would prohibit Chinese companies from training advanced AI models on remotely accessed AI chips? Or imposing tighter controls on remote access?
- Why has it still not closed a loophole it created that allows Chinese front companies outside China from making AI chips at TSMC or Samsung?
- Why has it not tightened controls on China’s access to semiconductor manufacturing equipment (which have not been updated in over 18 months - the longest the US has ever gone without updating them)?
- Why has it not imposed equivalent controls on all advanced AI models being served to China/Chinese companies?
- Why did it restrict access to all countries and foreign nationals accessing Mythos/Fable, not just China?

If the admin was serious about addressing the challenges posed by China in AI, it would be using export controls to address all of these questions and build a comprehensive strategy to prevent China from building or obtaining advanced models. But over the last 1.5 years, it has loosened or ignored controls on China, and only opened new loopholes in controls it inherited. If the admin truly has deep concerns about China’s access to advanced models, it has to act accordingly. It isn’t.

[Brian McGrail @brianmcgrail]

The "har har Anthropic got what it wanted!" takes are so stupid. Anthropic should not forfeit its right to object to all AI regulation because it called for a particular kind of AI regulation.

I know we are not the best at nuance in 2026 (particularly on this website) but come on.

(None of this is to say that Anthropic necessarily did the right thing here! Don't know enough facts. Just that these takes are dumb.)

[Brian McGrail @brianmcgrail]

@DavidSacks

[Brian McGrail @brianmcgrail]

@DavidSacks Do you agree with the Admin's decision to issue the export control?

[Charlie Bullock @CharlieBull0ck]

The AI governance writing that I’m least proud of is also the one that I find myself going back to the most. It’s the very first thing I wrote on AI, back when I was basically an intern at @law_ai_ in fall 2023.

The second section has a pretty detailed discussion of (a) how BIS can use deemed export controls and “is informed” letters to put together a de facto licensing regime for scary frontier models, and (b) potential first amendment objections to Commerce using deemed export authorities in that unprecedentedly aggressive way. At the time it seemed like a pretty speculative thing to write about, but now it seems like that analysis is going to be quite relevant in the coming months.

Shout out to @Christophkw for having a spooky ability to predict which AI policy topics are going to be relevant in three years and assign me to write about them

[dave kasten @David_Kasten]

Uhhh so incidentally, does anyone have a plan to prevent all the non-US citizen AI scientists from going to join foreign labs after they get bored of playing Wordle at work for a month, or are we just sort of planning on having the greatest counterproliferation failure since we deported Qian Xuesen in 1955 and gave Mao a rocket program?

[Brian McGrail @brianmcgrail]

For Fable/Mythos and AI regulation, this Churchill quote seems apt:

"Now this is not the end. It is not even the beginning of the end. But it is, perhaps, the end of the beginning."