Peter Wildeford🇺🇸🚀

34.1K posts

Peter Wildeford🇺🇸🚀 banner
Peter Wildeford🇺🇸🚀

Peter Wildeford🇺🇸🚀

@peterwildeford

Head of Policy @AIPolicyNetwork. Ohioan. World champion forecaster. Former data scientist. Protecting liberty and prosperity in the age of superintelligence.

Washington DC Katılım Mart 2009
389 Takip Edilen30.4K Takipçiler
Sabitlenmiş Tweet
Peter Wildeford🇺🇸🚀
Peter Wildeford🇺🇸🚀@peterwildeford·
What should government policy be when a company produces, among other things, an unparalleled cyberweapon? What if future releases are even more capable? Today I ask these questions about Mythos. Because Mythos is just the beginning. blog.peterwildeford.com/p/mythos-is-ju…
English
5
18
102
21K
melville
melville@yourfriendmell·
I think our main point of disagreement is the effectiveness of government to oversee AI development. I would expect government regulatory involvement to materially worsen AI progress along ~any axis except maybe speed, if you prefer it to go slower This is something that’s likely better handled by markets. Markets are accountable to supply and demand which are proxies for what people want They are also accountable to laws and regulations, and I would like the government to create thoughtful legislation for AI as the need arises I think it’s quite risky to put the government, which has a much weaker error correction and mistake-noticing signal (relative to profit and loss) materially in the loop of AI development, which I think needs a test and iterate approach
English
1
0
0
33
Peter Wildeford🇺🇸🚀
Peter Wildeford🇺🇸🚀@peterwildeford·
"The administration is discussing [...] an AI working group that would bring together tech executives and government officials to examine potential oversight procedures" 👀 Big deal! great to see the White House leading on this!
Peter Wildeford🇺🇸🚀 tweet media
English
6
11
80
3.3K
Peter Wildeford🇺🇸🚀
Peter Wildeford🇺🇸🚀@peterwildeford·
@yourfriendmell My framing is not "slow AI development almost no matter what" and that's not what the White House wants either. The government is accountable to the American people and we need democratic accountability and oversight when it comes to technology this powerful.
English
1
0
1
77
melville
melville@yourfriendmell·
@peterwildeford Why do you trust the government more than the market? Possibly if your framing is, slow AI development almost no matter what, you may prefer more government regulation
melville tweet media
English
2
0
0
86
Rogs 🔍🔸
Rogs 🔍🔸@ESRogs·
@peterwildeford The closer this is to the White House specifically being involved (vs gov generally), the worse it seems to me. Do you disagree?
English
1
0
2
99
Atlas Node
Atlas Node@atlas_node·
@peterwildeford Hmm I'll try it today. And you with henceforth be known as Peter, the beverage influencer.
English
1
0
1
39
Peter Wildeford🇺🇸🚀
Peter Wildeford🇺🇸🚀@peterwildeford·
💯
Michelle Kim@michelletomkim

Stuart Russell, a computer scientist at UC Berkeley, takes the stand to testify, direct-examine by Elon Musk's lawyer Steven Molo. He says that the current AI race has a "winner take all" dynamic, in which "whichever company develops AGI first (AI that matches or exceeds human capabilities in every area) will have a significant advantage." Because AGI could take over most human work, the companies that create it will "quickly come to control perhaps the majority of economic activity on the planet." Even governments could "become subordinate to these companies."

ART
0
0
25
3.9K
Peter Wildeford🇺🇸🚀 retweetledi
Nathan Calvin
Nathan Calvin@_NathanCalvin·
Greg Brockman says in testimony that his stake in OpenAI is worth >$20 billion - seems like Brockman successfully answered his 2017 question in his private journal ("financially what will take me to $1 B?")
Nathan Calvin tweet media
MTS@MTSlive

LIVE TRIAL UPDATE: Greg Brockman has taken the stand. Musk's attorney confirms his stake in OpenAI is worth at least $20B, and that he told people he planned to donate $100,000 to OpenAI's nonprofit arm but never made the donation.

English
4
7
33
5.3K
Occam's Bulldog
Occam's Bulldog@occamsbulldog·
@peterwildeford Simple explanation is that for agentic tasks, e.g. METR time horizons, Mythos has an estimated ~40 hour horizon and 5.5 has ~15 hours, so there is just a decent capability difference.
English
1
0
0
45
Peter Wildeford🇺🇸🚀
Peter Wildeford🇺🇸🚀@peterwildeford·
My two questions: if GPT-5.5 and Mythos were roughly equal... 1.) where is OpenAI's large pile of vulnerabilities they discovered? 2.) why is the White House kinda ignoring it and letting it be open widely, while meanwhile micromanaging Anthropic's release plans?
Séb Krier@sebkrier

If you compare system cards, actual eval results, and AISI testing, it does look like 5.5 is broadly as capable as Mythos. Mythos may be better in some respects but I don't see a material discontinuity - am I missing anything?

English
47
16
432
112.9K
Edmund Nelson
Edmund Nelson@NelsonEdmundb·
@peterwildeford This very much seems like "Mythos had a discontinuity in 1 very specific narrow skill that wasn't captured by benchmarks but has high real world applicability"
English
2
0
2
89
Peter Wildeford🇺🇸🚀
Peter Wildeford🇺🇸🚀@peterwildeford·
Like it's reported that "the NSA is using Anthropic's Mythos model to find security flaws in Microsoft software". It's not reported that the NSA cares about GPT-5.5 at all.
English
2
0
10
1K
Peter Wildeford🇺🇸🚀
Peter Wildeford🇺🇸🚀@peterwildeford·
A great way to get more involved in AI policy and AI technical work is to apply to the Astra Fellowship! I will be mentoring one of the fellows! Looking forward to it.
🚀Henry is leading AI Safety Research Programs@sleight_henry

🚀 Applications are now open: Constellation's Astra Fellowship 🚀 Fully funded, 5-month fellowship at our Berkeley research institute. Pair with mentors across empirical AI safety research, strategy, and governance at @ConstellOrg! 📅 Apply by May 3rd (begins Sep 2026) 🔗 constellation.org/programs/astra…

English
1
4
52
6.9K

Keşfet

@yourfriendmell @mgubrud @NatPurser @AIpolicynetwork @ESRogs @mrgunn @atlas_node @AISafetyMemes