Samuel Rees (@samuelcrees) header
Samuel Rees (@samuelcrees) profile picture

Freelance Audio Engineer. Software Developer at @helicoptersofDC for the CopterSpotter. 703-967-2230 on Signal.

Washington, D.C.
Joined March 2010
Photo by Samuel Rees (@samuelcrees)Photo by Samuel Rees (@samuelcrees)Photo by Samuel Rees (@samuelcrees)Photo by Samuel Rees (@samuelcrees)Photo by Samuel Rees (@samuelcrees)Photo by Samuel Rees (@samuelcrees)Photo by Samuel Rees (@samuelcrees)Photo by Samuel Rees (@samuelcrees)Photo by Samuel Rees (@samuelcrees)Photo by Samuel Rees (@samuelcrees)
Samuel Rees retweeted
For the first time, we’ve given external researchers a way to study AI’s impacts using real, privacy-preserved Claude usage data. To date, this work has only been possible within AI labs. We can’t tell the whole story alone, so we opened up our tools. anthropic.com/research/enabl…
200
235
99
2,259
558,173
Samuel Rees retweeted
The government can already legally buy your location data, browsing history, and social media activity without a warrant. The only thing that prevented mass surveillance from that data was the inability to process it all. LLMs fix that. “All lawful purposes” includes this.
Dario’s main claim is that laws have *not* kept pace with AI use because of its ability to sort through mountains of data, which has been nearly impossible until recently. Therefore, the reason Anthropic specifically points to is not mass surveillance, but rather “all lawful purposes.” Due to the law's limited scope (which doesn't fully account for LLMs), we’ve been effectively saved from mass surveillance by the lack of technology that can make sense of mountains of data. That is no longer the case, which is why Anthropic was so adamant about adding the phrase “no mass surveillance” rather than “all lawful purposes.” In other words, it seems likely that using LLMs to filter through mountains of surveillance data falls within the purview of the law, and Anthropic did not want to remove its clause because it knew the law had not adapted to our new reality. So, it seems to me that in OpenAI’s case, it’s partially dependent on their ability to detect when the DoW is using their models to analyze all the mass surveillance data. They will have a “safety stack” for the models, but will the models really block reading through data for mass surveillance? Between API calls, it may look incredibly innocuous compared to biosecurity. More importantly, the kind of work the DoW is trying to do *may* actually be completely within the law, but it still significantly enhances their ability to conduct mass surveillance under the current state of the law. With that in mind, how will OpenAI avoid getting sued by the DoW when they ask OpenAI to do the “lawful actions” that significantly help with mass surveillance? At that point, OpenAI has three options. They comply, which means the safety stack isn’t a real guardrail. They refuse, and the DoW sues for breach of contract because the lead clause of the contract they signed grants “all lawful purposes,” and the action is lawful. Or they refuse, don’t get sued, but get the Anthropic treatment: a supply chain risk designation, a federal ban, and public attacks from the president. It seems that if it goes to court, they will almost certainly lose. So, what happens if they are pressured?
9
99
6
509
31,148
OpenAI has released the language in their contract with the DoW, and it's exactly as Anthropic was claiming: "legalese that would allow those safeguards to be disregarded at will". Note: the first paragraph doesn't say "no autonomous weapons"! It says "AI can't control autonomous weapons as long as existing law (that doesn't exist) or the DoD says so." Similarly, the mass surveillance use cases will "comply with existing law", but many forms of data collection that we'd consider "mass surveillance" are things that the NSA has consistently argued are legal under current law.
119
1,520
125
6,240
413,772
Replying to @sama
What do you think about this community note? Is it correct?
9
11
1
479
33,919
If I worked at OpenAI, I would want to see the contract details, or have a critical mass of employees from a range of teams/perspectives see the contract details. What has been reported so far is confusing + the details matter. xitter.gq/axios/status/202753251…
NEW: The Pentagon has agreed to OpenAI's rules for deploying its technology safely in classified settings, though no contract has been signed, a source tells Axios. The department appears to have accepted conditions similar to those put forth by Anthropic. axios.com/2026/02/27/pentago…
16
35
4
591
62,444
Samuel Rees retweeted
It’s extremely good that Anthropic has not backed down, and it’s siginficant that OpenAI has taken a similar stance. In the future, there will be much more challenging situations of this nature, and it will be critical for the relevant leaders to rise up to the occasion, for fierce competitors to put their differences aside. Good to see that happen today.
1,417
2,476
430
25,607
3,299,022
Replying to @usairforce
@usairforce is 5:30am on a Sunday really the best time for 1st Helicopter Squadron to joyride 5 ships at low altitude over the city? Hope there’s a reason you have ADS-B off cause I’ll be FOIA’ing. cc: @SenTedCruz @SenatorCantwell
8
8
6
59
13,053
It’s not the sound of freedom.
3
7
40
3,288
CopterSpotter Update: Tagged helicopter spots now show flight paths! Tag spots by selecting a live aircraft or browsing nearby flights — even from the past. Link below and in bio 🚁🎉
1
3
15
4,627
Tesla really made the best gear selector in the world only to discontinue it
97
14
7
496
72,950
Air Force 1st Heli Squadron, which typically used ADS-B transponder once a year, is currently flying 4 helicopters with ADS-B on. Could this be because of the urging of @SenatorCantwell at last weeks senate hearing? Following up on the letter @EleanorNorton wrote at my request?
3
2
11
1,124
Replying to @CaptMarkKelly @UAW
Extremely disappointed that you got rid of your EV only to purchase a gasoline vehicle.
20
2
189
15,326
Musk campaigning: There should be no need for FOIA, because all government records should be public by default. Musk governing: My “efficiency” team isn’t subject to FOIA and if we are, we’ll need three years to produce records. nytimes.com/2025/03/10/us/po…
There should be no need for FOIA requests. All government data should be default public for maximum transparency.
2
17
27
9,332
Yesterday republicans voted to raise the deficit by $4 trillion dollars and today @DOGE is fretting about 107 unused photoshop licenses. Y'all got played.
351
2,634
50
36,617
873,018
Please. No more speeches in the Senate about “freedom” until one Republican has the guts to stand up to Trump’s lies about Ukraine & Putin. Ukraine did not start this horrific war. Russia did. Zelensky isn't a dictator. Putin is. The US must not enter into an alliance with Putin.
7,448
15,349
700
108,397
2,781,953
I would trust DOGE more if they started at the top with the $64 billion in PPP fraud instead of instantly nuking humanitarian efforts for babies in Africa. who does that.
7,337
6,662
465
91,423
3,580,451
how it started, how it's goin #Musk
3,681
15,000
685
147,402
3,887,248
Samuel Rees retweeted
"we're spending $1 million in taxpayer money so professors can mail each other little notes?? we have a post office—these toys for academics are emblems of waste" —DOGE, 1969, killing arpanet
I'm not sure why it's so hard for so many smart people to understand what's going on here The crazy science projects are emblems of the waste, they are not the point of the efficiency thrust
114
1,327
43
17,347
1,075,156
.@samuelcrees has worked tirelessly to bring live helicopter locations to our map on CopterSpotter.com. This gives residents the best possible insight into the actual purpose of our region's helicopters, not simply who owns them. patreon.com/posts/live-helic…
1
1
1
3,512