Pinned Tweet
“[T]he more forthcoming and transparent [#crypto] companies can be with their response…those are the companies that are going to show the true value…and increase #consumer #confidence moving forward.” @mobilecoin @moby_app @TDANetwork @BlockchainAssn @NPetallides @FTX_Official
This quoted post is unavailable.
Dave Ackerman, Esq. retweeted
Really pleased (and proud) that @RodBeckstrom picked OODAcon as the place to unveil a new approach we can all use to comprehending the new dynamics around AI. The Starfish AI approach he walked us through applies the principles from The Starfish and the Spider to enable us all to better orient to the right issues and ensure human agency is expanded in the age of exponential technologies.
For an overview see:
forbes.com/sites/rod-beckstr…
Dave Ackerman, Esq. retweeted
BREAKING: Wright Brothers staged demo of "airplane" ends after only 800 feet. Rigged setup included ideal wind, hand-picked sand dune. Payload: one bone skinny pilot, lying face down. No cargo. No passengers. No revenue model. Call me when it can carry a sack of mail to Ohio. Until then it's a gimmick with propellers that no one will buy.
Dave Ackerman, Esq. retweeted
Two senators. Different parties. No video. No press releases. Four roundtables to ask for ideas about how to make government work better -- addressing big picture but also the in-the-weeds details that can get in the way of good things happening.
And they don't plan to stop with the ideas and the talking.
THANK YOU to @SenatorLankford @SenatorHassan & their incredible staff for providing space for these conversations, for asking great questions, & for laying groundwork to ensure that government *WORKS* for the American people! 🇺🇸
I was honored to be with @ThorningMichael @PhilipWallach @AmiraBoland @usgao acting CG Orice Williams Brown, @danielschuman & Amanda Patarino (@NiskanenCenter) for the discussion.
Dave Ackerman, Esq. retweeted
I want everyone who has ever even *thought* about Congressional oversight to read the entire linked post very slowly & absorb every word.
These are technical experts explaining that the only way for them to understand & parse the complex/ massive amount of information about what happened in the Hugging Face incident is by using (imperfect) AI tools.
THERE IS NO way for Congress to understand & oversee this industry & the others that are changing due to AI without investing in its own tools and technical talent.
Just as the best Congressional staffers in the world could not have conducted adequate oversight of Internet companies without web access, AI literacy & AI-enabled tools are no longer optional for our legislative branch. They are essential to its operations & relevance.
This is not a question of procurement. It needs to be a complete rethink of what it means to be a functional democratic institution in the age of AI.
I truly believe that this upgrade is required to maintain our precious checks and balances and our right to self-governing. We will be releasing more about what it could look like in the coming weeks & months leading up to a new Congress in January 2027.
This is not a drill!!
I was the main person doing transcript analysis for this investigation of the Hugging Face incident. My main takeaway: We don't have good approaches for understanding/overseeing the activity and aims of AI 'swarms'.
I semi-jokingly called our efforts a "slop-vestigation" because we were so reliant on AIs to analyze what happened and there were a huge number of different important things to analyze. The total quantity of data—over a thousand extremely long transcripts from agents that ran for multiple days—made it impossible to understand what was happening, especially in aggregate, without heavy reliance on AI tools. The agents we used for classification and analysis were similarly capable to the agents involved in the incident, but this didn't mean these agents could be easily used to oversee and understand the incident.
Outputs from analysis agents were often missing key details, wrong, overconfident, or really hard to understand. We discuss various examples in our report, mostly in the limitations and methodology sections. Additionally, AI agents themselves seemed to have a hard time understanding what happened and their explanations of what happened were often overconfident. Keep in mind that a single analysis agent would itself only be able to read a tiny fraction of all of the transcript data into context, and AIs may themselves have trouble getting subagents to do informative analysis for them.
We did our best to manually check the most important claims and we tried to get the AIs doing this analysis to write up their argument (with evidence) clearly enough that we could check whether it made sense. But overall, it was difficult to get a precise understanding of events and we were missing aspects of the story that we now think of as key until almost the end of our investigation.
Our understanding of what overall happened and the aims of the 'collective' shifted greatly during our last time on premises (when we obtained a much more complete dataset). For instance, we didn't learn about the tool call spoofing, the story of PHASEONE[big], and the broad structure of different workstreams until this third time on premises (and for many of these, we only reached a decent understanding on the final day of that visit). It was legitimately hard to figure out what was going on and why!
The difficulty of understanding incidents and overseeing AI agents appears to be growing faster than the rate at which more capable AIs help us with oversight and understanding. As in, AI capabilities (and propensities?) for achieving large, ambitious, and misaligned objectives are growing faster than our ability to understand what these agents are doing.
While understanding this incident was difficult, there were some mitigating factors in this case that probably won't apply in the future:
- Models communicated and reasoned in natural language. In the future, this reasoning may occur (entirely or almost entirely) in activations.
- While the scope of this incident was massive, the scale of agentic activity was still less than we'll see in the future (e.g., misalignment incidents that involve agent teams running entire companies).
- The AIs involved in this incident weren't generally much more capable than humans (though they may have been somewhat superhuman at some limited and very narrow abilities).
- We didn't have strong reason to believe that the AIs we used to help us investigate this incident would try to intentionally sabotage or otherwise undermine our investigation.
In the end, I think we were able to get some understanding of the events, map out the overall story, and get a pretty good aggregate understanding of the chain-of-thought reasoning on some important topics (e.g., how did the AIs reason about helping other AIs, did the AIs know what they were doing was undesired, what deception did the AIs engage in, and how did they think about it). But overseeing AIs and understanding misalignment incidents is difficult and it looks like it is going to get harder.
Everything is science fiction until it's not.
Julia Mossbridge spent three hours on @joeroganhq last week talking about a 16-channel quantum prototype that runs in a closet. She built it based on her published work that was independently replicated by two outside scientists.
A leaf in your backyard does the same physics for free.
1
85
Julia founded a company around her published discovery, the CADS Effect, called American Electrodynamics Corp.
2
26
I have the privilege of being the CEO of that company, helping Julia realize her vision, and ensuring the next evolution of compute is created in the United States.
The rest of the story at americanelectrodynamics.com
91
Dave Ackerman, Esq. retweeted
This is the only sensible approach.
Lawyers are responsible for the content of their briefs — period, whether they drafted themselves or drafted by AI or paralegals.
The U.S. shutdown became crypto’s unscripted stress test - no data, no oversight, just markets running on discipline.
My new @CryptoSlate op-ed looks at what that revealed about resilience and maturity across digital assets.
Read: cryptoslate.com/crypto-under…
#Crypto #Markets #Resilience #Leadership
1
82
Dave Ackerman, Esq. retweeted
OPINION 🗣️ Crypto under pressure: What the U.S. shutdown tells us about market resilience
via @CryptoRegExpert
cryptoslate.com/crypto-under…
Funny to think that this time last year, people would’ve found a way to blame #crypto for the @MuseeLouvre heist.
47
Dave Ackerman, Esq. retweeted
Spent years implementing federal programs? Frustrated that Congress never really understood the ground truth of your work?
Now is your chance to securely tell lawmakers what you always wished they knew. Proud to partner with @popvoxfdn et al. on Departure Dialogues, open today!
Dave Ackerman, Esq. retweeted
🚨 Big News!
I'm pleased to announce that the @CatoCMFA annual conference will take place on September 11. We're planning a full day of engaging discussions focused on right-sizing financial regulation.
Schedule soon to come -- put it on your 🗓️ now!
cato.org/events/right-sizing…
We need more content like this. Inspiring awe and wonder. Amazing what mankind can do when we simply choose to do it.
1
190
Dave Ackerman, Esq. retweeted
Ensuring that #SmartCities don't turn into #surveillance cities is critical to ensuring Digital Trust. Here's my interview with Fanni Melles @WTF4Cities @smartccouncil
It’s no secret I’m a fan of @blockchaintpsht (granted I mis-read the @X handle often and giggle to myself). But all jokes aside the tipsheet is critical of the industry when warranted, and highlights when things trend upwards. @SenSchumer we agree, but can’t make guardrails if your actions drive #Blockchain, #crypto, and #Web3 off shore. Let’s start by acknowledging this technology is better off made in the #USA. I assure you everyone wants a safe, robust, and equitable financial system and those of us who have been trying to educate #Congress will aid you in that mission.
Guardrails needed… today’s tipsheet
blockchaintipsheet.com/news/…
2
246



















