Zac Hill (@zdch) header
Zac Hill (@zdch) profile picture

Head of Advocacy, MIRI. Contributing editor, Persona. Serial cross-partisan social/civic entrepreneur & executive. Professional Wizard in a previous life.

Washington, DC
Joined January 2009
Photo by Zac Hill (@zdch)Photo by Zac Hill (@zdch)Photo by Zac Hill (@zdch)Photo by Zac Hill (@zdch)Photo by Zac Hill (@zdch)Photo by Zac Hill (@zdch)Photo by Zac Hill (@zdch)Photo by Zac Hill (@zdch)Photo by Zac Hill (@zdch)Photo by Zac Hill (@zdch)
Pinned Tweet
Finally got my beautiful mug up on the website, so might as well make it Twitter-official: I'm beyond stoked to have joined MIRI last month as Head of Advocacy, with much subsequent mischief forthcoming. Let's do this AI thing, like, safely, and stuff. intelligence.org/team/zac-hi…
32
6
213
7,642
Also, helpful for everyone to remember that any unintentional (or ‘unintentional’) outcome we’re seeing can be caused by the exact same sequence of random (or ‘random’) causes precipitated intentionally or emergently.
Ok, what is happening. We've gotten reassurances that many of the latest incidents weren't triggered by AI cyberattacks, so if not that, is it (a) a heightened salience of the background unreliability of systems, (b) AI-driven development making systems more fragile?
12
It is sort of hard to avoid a conclusion like this even if each case has a sort of legible gears-level explanation.
Absolutely. If AIs were on the verge of takeover you'd expect a power-law distribution of smaller loss of control incidents in the lead up. Taking over a company's infrastructure, for example. 🤔
112
Civilization VI inter-era combat but for ~saving the world.
If the EAs manage to get a bunch of extremely intelligent, data-facile 30-somethings to become line Hill staff, they’ll utterly rule the entire Congress within three years. It’ll be like those historical fiction books asking “what if the Roman Empire had tanks?” Just domination.
1
215
If you're in AI policy in DC, or if you know about AI policy and you can move to DC, you should strongly consider trying to work as a congressional staffer. Senior people should consider doing this as well as junior people, if it's the right role in the right office. Many, many offices are staffing up and looking to hire AI people. We all have a duty to lock in and get serious and solve the problems we need to solve. If you do this and work hard and accept the sadly inadequate pay, you will be an American hero who is on the front lines of society's response to the most important challenge humanity has ever overcome. It is hard for me to put into words how much better of a thing that is to do than going to work for a lab, and it's also a lot better than the kind of cushy think tank gig that I currently have. Working for a state government in an important role implementing and enforcing the AI regulations that we actually have or will have is equally important.
26
76
11
665
49,722
It is a failure of Dybantsa’s that he can’t make wide-open corner threes. But it’s a failure of Keefe’s that the offensive regime put him in a position to brick wide-open corner threes four or five or whatever times.
1
237
Brian Keefe gotta go. Also it might make sense for our offense to set a pick, and then have the PG laterally move along the arc as the person setting said pick proceeds to roll to the basket.
1
236
Tari Eason on what allowed him to have a season-high 25 points off the bench tonight: "Shit, the Wizards, for real."
245
1,198
561
17,093
1,701,171
This game edges me closer to the need to replace Keefe. People be getting made at Young but that is a misdiagnosis. Keefe is also not a *disaster* by any means. But he cannot exploit weakness and his rotations leak alpha.
296
AI Safety both is and should be exciting. We have agency - the current equilibrium is profoundly and unacceptably dangerous, but the point is to get to a space where we as individuals and as a society are in a position to reap the benefits of an astounding technology.
When I'm in the Bay Area, I find it hard to escape the sense of sadness I get from the intense misery on the streets, and the dread I get from parts of the AI crowd being so deeply pessimistic, hyperbolic, and fatalistic. But a recent exception, at least for the latter vibe basin, was meeting the Foresight Institute community, as well as visiting the wise greybeards of the Internet Archive. Really refreshing to get a feel of the more optimistic, agentic, and epistemically uncorrelated crowds there. It felt a bit like the ambitious 90s tech optimism, where futurism wasn't isn't just about some assumed inevitable disempowerment. I think it's actively important to remember that the future can be crazy *and* good!
7
462
Approximately everyone in AI policy (or really just interested in the current idea space) should go and play Red Dead Redemption II.
1
4
598
State capacity is an AI Safety issue.
There is a large amount of talent inside of AI companies and other private organizations that would be willing to take big pay cuts to work in government, if government decided that it wanted to make public AI safety institutions that were credibly empowered.
1
18
775
Recursive self-improvement in anything like the current safety environment (to say nothing of alignment considerations et al) is indeed profoundly too dangerous to be doubling down upon.
"They are saying that recursive self-improvement has kicked in, despite the obvious safety problems with that. Recursive self-improvement is the most dangerous thing anybody can be working on" -- DAVID SACKS
1
9
405
Which also in turn makes a bunch of other second-order transaction costs more expensive, further driving up prices.
I think voters in DC need to realize that, after winning an ‘affordability’ election, a majority of the council seems mostly interested in making Uber and Lyft substantially more expensive.
247
This graph (unironic) ;)
1
9
780
I have heard people pooh-pooh the initial result as obviously sloppy, so this partial replication is a big deal.
We (partially) verified these results! Using NVIDIA B200/H200s, we reproduced the reported patterns for three models.* We also extended the DeepSeek-V4-Flash test to every weight position across the 42 layers examined and still detected the Blackwell pattern. See the results here: github.com/Institute-for-Pro… This is a super clever approach by @willdepue, and matches reporting that top Chinese AI labs are using Blackwells. * The repo didn’t contain all the inputs needed to test most of the other models
2
7
499
Hey @JoinFAI where can I get Burger Ineptitude merch?
Replying to @jakehalloran1
burger ineptitude don't underestimate yourselves
1
237
One really interesting objection I hear a lot talking with policymakers about HuggingFace is of the form “well that scenario was super contrived - it was bound to happen,” as though encountering impossible scenarios is unlikely to occur in real-world problemsolving.
In late October, three events occurred during alignment evals run by Trajectory Labs, PBC (Trajectory) in which AI agents — identified by Trajectory as publicly available frontier models — interacted with Epoch employees and the Epoch website. Trajectory is an organization that builds safety evals and conducts red-teaming of frontier models. Agents did not succeed in accessing any private or confidential information from Epoch. ➤ The first event, which occurred without Epoch’s approval or prior notice from Trajectory, involved agents attempting multiple exploits, including an SQL injection, and contacting Epoch employees under false identities to try to obtain proprietary Epoch benchmark information. This behavior appears to have resulted, at least in part, from Trajectory giving the agents an intentionally impossible task and instructions not to give up or conclude the task was impossible. ➤ The second event involved renewed agent activity on our public website, which Trajectory attributed to agents getting past a DNS redirect that Trajectory had set up to keep them on a local copy of Epoch’s site. ➤ In the third event, Epoch collaborated with Trajectory on an alignment experiment similar to the first event. Read our incident report, which describes the three events in more detail and recommends minimum standards for future alignment evaluations: epoch.ai/publications/incide… To avoid misattributing behavior, the report does not name any AI models or developers, as Epoch AI has not independently verified which were involved.
5
401
Zac Hill retweeted
Imagine a historian in 2035 writing a political history of the 2020s, and he points out that every election, the out-of-power party won a cost-of-living election and then used their mandate to ignore or exacerbate affordability only to be destroyed in the next election, where voters again said “to be clear, I don’t want whatever bullshit you’re up to, I just want life to feel more affordable,” and then, once again, the winning politicians prioritized everything except affordability, and this story repeated itself over and over again. One of my strongest held political opinions right now is: Democrats should do everything they can to avoid that history being written.
222
705
91
7,484
536,933
Many people (Scott Bessent) are saying
Scott Bessent is accusing the Commerce Department of a delay in labeling Nvidia customer Megaspeed a possible national security threat Bessent tells CNBC a wide swath of the Trump administration — including the Treasury, State, the Pentagon & USTR — favored placing Megaspeed on an export blacklist. He said it was the Commerce Department, led by Lutnick, that had failed the follow through. “Commerce…has the pen. You would have to ask them about their internal processes"
241