Lightnews — Scholar-powered news

Garrison Lovely @garrisonlovely.bsky.social · 1d

And if you want to stay up to date with what's happening, subscribe to Obsolete (will be more active once I finish my book in November): substack-proxy.glitch.me/articles/ob...

Obsolete | Garrison Lovely | Substack

Reporting and analysis on capitalism, great power competition, and the race to build machine superintelligence from journalist w/ work in NYT, Nature, BBC, TIME, and more. Click to read Obsolete, by Garrison Lovely, a Substack publication.

substack-proxy.glitch.me

1

Garrison Lovely @garrisonlovely.bsky.social · 1d

If you want to talk, I'm on Signal at: Garrison.06

Props to Josh for speaking out and for taking seriously the gravity of what OpenAI is trying to do.

1

Garrison Lovely @garrisonlovely.bsky.social · 1d

If you work there, I really encourage you to familiarize yourself with what your bosses are doing. Here are some places to get started:

www.wsj.com/politics/si...
www.newyorker.com/magazine/20...
www.obsolete.pub/p/exclusive...
www.obsolete.pub/p/sam-altma...
www.obsolete.pub/p/inside-te...

Inside Tech's Risky Gamble to Kill State AI Regulations for a Decade

Republicans slipped a controversial provision into the “One Big Beautiful Bill” — now facing bipartisan backlash and internal party rebellion

www.obsolete.pub

1

Garrison Lovely @garrisonlovely.bsky.social · 1d

For instance, believing that OAI really does want regulation, but just wants it to happen at the federal level — despite the fact that OAI publicly called for preemption of state-level bills with no binding replacement, something that has literally never happened before.

1

Garrison Lovely @garrisonlovely.bsky.social · 1d

Wow, OpenAI's head of mission alignment just spoke out against the way the company has been using subpoenas to intimidate and disrupt political opponents.

A surprising number of OAI rank & file have no idea what their leadership is doing to kill regulation.
x.com/jachiam0/st...

1 4

Garrison Lovely @garrisonlovely.bsky.social · 3d

Full context at time stamp: youtu.be/rF0tQtDMwHM...

If you enjoyed this, subscribe to Obsolete, where I write on the economics and geopolitics of the race to build machine superintelligence: substack-proxy.glitch.me/articles/ob...

Obsolete | Garrison Lovely | Substack

Reporting and analysis on capitalism, great power competition, and the race to build machine superintelligence from journalist w/ work in NYT, Nature, BBC, TIME, and more. Click to read Obsolete, by Garrison Lovely, a Substack publication.

substack-proxy.glitch.me

Garrison Lovely @garrisonlovely.bsky.social · 3d

Would have been nice if the interviewer asked him about these moves, but alas.

1

Garrison Lovely @garrisonlovely.bsky.social · 3d

AND more likely than not to be developed during Trump's presidency.

OpenAI execs are teaming up with Andreessen Horowitz to spin up a massive AI super PAC, modeled after the crypto playbook of intimidating congress into supporting its agenda. www.wsj.com/politics/si...

1

Garrison Lovely @garrisonlovely.bsky.social · 3d

It's also openly called for the federal govt to preempt state-level AI regulations with no binding replacement — an unprecedented move and not something you'd expect from someone who thinks that superintelligence is the biggest threat to humanity
api.omarshehata.me/substack-pr...

1

Garrison Lovely @garrisonlovely.bsky.social · 3d

But it's also hard to reconcile this view with the increasingly intense anti-regulation position OpenAI is taking. Altman clarifies that there's a diff between regs on x-risk and regs on banking software, but OAI opposed SB 1047, which was focused on catastrophic risk.

1

Garrison Lovely @garrisonlovely.bsky.social · 3d

Well, now I need to update my book.

To my knowledge, this is the first time Sam Altman hasn't downplayed or dismissed AI existential risk since early 2023.

TBC, I think it's good of Altman to say this if that's what he actually believes...
x.com/ai_ctrl/sta...

1

Garrison Lovely @garrisonlovely.bsky.social · 4d

Original screenshot from this article: www.computerweekly.com/news/366629...

Autocratic UAE gets democratic artificial intelligence | Computer W...

US-China rivalry has come to a head in authoritarian Gulf state UAE, where OpenAI is attempting to make a stand for democracy and free speech.

www.computerweekly.com

Garrison Lovely @garrisonlovely.bsky.social · 4d

I wrote about this in my Current Affairs essay on McKinsey: www.currentaffairs.org/news/2019/0...

1

Garrison Lovely @garrisonlovely.bsky.social · 4d

This reminds me of arguments that McKinsey would make to justify working for Gulf autocracies. However, academic research has found that the opposite tends to happen: companies abandon human rights to conform to their wealthy clients.
x.com/ShakeelHash...

1

Garrison Lovely @garrisonlovely.bsky.social · 11d

Oh and Daron said he was very excited for my book, so you should be too!

Garrison Lovely @garrisonlovely.bsky.social · 11d

Be the first to know when the book is available by signing up here:
www.obsoletebook.org/

And to stay up to date with my journalism, subscribe to Obsolete:
substack-proxy.glitch.me/articles/ob...

Obsolete | Garrison Lovely | Substack

Reporting and analysis on capitalism, great power competition, and the race to build machine superintelligence from journalist w/ work in NYT, Nature, BBC, TIME, and more. Click to read Obsolete, by Garrison Lovely, a Substack publication.

substack-proxy.glitch.me

1

Garrison Lovely @garrisonlovely.bsky.social · 11d

Daron and Sandhini were a pleasure to work with, as was everyone from the Nobel Foundation, the Swedish Consulate, and the Astralis Foundation.

1

Garrison Lovely @garrisonlovely.bsky.social · 11d

Preparing for the conversation actually helped me crack the thesis of my book, so if nothing else comes of it, I'll still be grateful!

The event was not recorded, but we were lucky to have a stacked audience (e.g. another Nobel-winner for the discovery of mRNA was in attendance)

1

Garrison Lovely @garrisonlovely.bsky.social · 11d

I got a chance to push each of them a bit. Daron on his skepticism of AI capabilities+pace of progress. Sandhini on the sufficiency of self-governance as AI grows more capable+ubiquitous.

I was surprised by each of them, and I'll have more to say in my book.

1

Garrison Lovely @garrisonlovely.bsky.social · 11d

Last week, I had the incredible privilege of moderating a conversation on AI+geopolitics w/ Daron Acemoglu & Sandhini Agarwal (OpenAI trustworthy AI lead) for the Nobel Foundation at the Swedish Consulate.

It was lively+substantive (one audience member was pleasantly surprised)
x.com/swedennewyo...

1

Garrison Lovely @garrisonlovely.bsky.social · 12d

Newsom just signed Sen. Scott Wiener's SB 53, exactly one year after he vetoed SB 1047. www.gov.ca.gov/2025/09/29/...

Governor Newsom signs SB 53, advancing California’s world-leading artificial intelligence industry | Governor of California

www.gov.ca.gov

Garrison Lovely @garrisonlovely.bsky.social · 16d

(GPT-5 and Gemini 2.5 were far more useful.) Anyway, here are the links if you wanna try it for yourself openai.com/index/gdpval/
economics.mit.edu/sites/defau...

1

Garrison Lovely @garrisonlovely.bsky.social · 16d

Funniest part of all this is that I was asking models how OpenAI's new GDPval results would affect Daron Acemoglu's GDP findings from his 2024 paper. GDPVal found Opus 4.1 was the clear winner at these tasks, but it gave me the clearly worst answer to my question

1

Garrison Lovely @garrisonlovely.bsky.social · 16d

"The human"

1 2

Garrison Lovely @garrisonlovely.bsky.social · 17d

If you enjoyed this, subscribe to Obsolete, where I write on the economics and geopolitics of the race to build machine superintelligence: www.obsolete.pub

Obsolete | Garrison Lovely | Substack

Reporting and analysis on capitalism, great power competition, and the race to build machine superintelligence from journalist w/ work in NYT, Nature, BBC, TIME, and more. Click to read Obsolete, by G...

substack-proxy.glitch.me