The “free speech” and “cancel culture” panics have led to overt state censorship because that was always their purpose. (Gift link) www.theatlantic.com/ideas/2025/1...

"we need to start to use a percentage of the balance sheets of these companies in order to benefit as many americans as possible." perhaps an interesting, if inadvertent, admission

people certainly (probably rightly) think that AI/electricity costs will be a top issue for voters in 2026. so far under-appreciated: the role AI will play in making Medicaid work requirement determinations, potentially kicking tens of thousands off their benefits

“In America, one can never escape the inevitability of race. It’s happening now, as you read this—that inevitability, the one built on old power imbalances and a racial hierarchy, is pushing industries to the brink” reckoning with the reality of DEI for @wired.com www.wired.com/story/dei-di...

new from me: are data centers going to wipe out water supplies? how much water does ChatGPT really use? and what's going on with that big correction in Empire of AI? i went long on the conversation around AI and water, and how it's actually about what we want resources to be used for:

SFO ✈️ DCA

feeling the AGI

DCA ✈️ SFO

free advice for democratic policymakers: if the president is about to sign an unlawful executive order attempting to preempt every state AI policy by directing the DOJ to sue states, perhaps don't come out of the gate with "we are not against preemption, we just want preemption lite"

free advice for democratic policymakers: if the president is about to sign an unlawful executive order attempting to preempt every state AI policy by directing the DOJ to sue states, perhaps don't come out of the gate with "we are not against preemption, we just want preemption lite"

"We tend to imagine corruption as ... money buying votes, quid pro quos in backrooms. But money's real power is quieter and deeper. It decides which candidates get to run, which policies are thinkable, and whose voices get amplified or ignored." data4democracy.substack.com/p/money-does...

so I understand it: we have to preempt every state AI law because we have to beat China. simultaneously, we have to provide China access to some of our most advanced chips, in order to beat China.

I was told by the much vaunted AI Action Plan this wouldn’t happen! …almost as if the AI Action Plan wasn’t worth the paper it was written on when in service of an authoritarian regime

"we want to financialize everything and create a tradable asset out of any difference in opinion” is perhaps the apotheosis of learning the wrong lessons from the past two decades.

we can be confident that the large technology companies that outsourced a lot of this work will work tirelessly to stand up for their international contractors and workers! [taps ear] www.reuters.com/world/us/tru...

nearly 1,000 people stood in line this week for a chance at emergency rental aid in DC on the *one day* they could apply after the phone lines crashed due to overwhelming demand www.washingtonpost.com/dc-md-va/202...

proponents of a state AI policy moratorium often argue that state proposals aimed at combatting algorithmic discrimination are “redundant” because of existing civil rights law. this view is very common! but it is flatly wrong.

the newest report from my colleagues at Upturn interrogates a set of under-examined automated systems that directly impact the material conditions of your friends, family, and community: Medicaid home care eligibility algorithms.

I have yet to see someone clearly answer why a state AI policy moratorium would not imperil basic state regulation of insurance underwriting and pricing

CDG ✈️ JFK ✈️ DCA

ARN ✈️ CDG

found the frontier

Malmö 🚅 Stockholm

literally hirevue had to discontinue this service in 2021

Copenhagen 🚅 Malmö

IAD ✈️ CDG ✈️ CPH

the newest report from my colleagues at Upturn interrogates a set of under-examined automated systems that directly impact the material conditions of your friends, family, and community: Medicaid home care eligibility algorithms.

3/ There’s no escaping the fact that these automated Medicaid eligibility determinations are arbitrary and subjective: based on our analysis, someone who is eligible for home care in Mississippi and Nebraska could be *ineligible* in Missouri, New Jersey, and DC.

2/ Our report offers an in-depth analysis of five states’ eligibility algorithms, case studies on how states work with vendors to develop their algorithms, and historical research on the emergence of the federally mandated eligibility standard for Medicaid home care.

1/ NEW: Read our new report, “Calculated Need: Algorithms and the Fight for Medicaid Home Care.” We show how automated eligibility determinations are political tools that draw a subjective line between who is deserving and undeserving of getting care in their home. www.upturn.org/work/calcula...

putting other methods issues aside, in a world of algorithmic feeds, why would *social media following* be the best proxy for “attention”???

Salesforce told ICE that its A.I. software could help the agency nearly triple its staff, according to internal documents, as President Trump expands immigration raids and deportations around the U.S. www.nytimes.com/2025/10/16/u...

Our longtime Executive Director @harlanyu.bsky.social announced today his plans to depart Upturn in 2026. Over 15 years, we’ve built a strong, principled & impactful organization fighting for racial, economic & social justice through a tech lens. We’re excited for Harlan and will miss him dearly!

learned how to ride a bike today

DEN ✈️ DFW ✈️ DCA

DCA ✈️ DFW ✈️ DEN

Not sure what’s most alarming fact, that the administration doesn’t appear to know what the boat it blew up was carrying or where it was headed, that they haven’t identified a legal authority for the action, or that it’s the first in a series of planned attacks. www.nytimes.com/2025/09/03/u...

This is my biggest gripe with the whole field of AI safety. Safety people are strangely uninterested in ethics (different area) and are completely uninterested in, say, a currently deployed regression-based system doing bad stuff in hospitals. They are exclusively interested in sci-fi hypotheticals.

ChatGPT appears to have fueled an individuals paranoia that everyone was out to get him, including his mother. It appears to have resulted in a murder suicide. www.wsj.com/tech/ai/chat...

@beenwrekt.bsky.social on the venal negligence of "AI Safety" www.argmin.net/p/the-banal-...

same week

these excerpts from the lawsuit against OpenAI are just so profoundly upsetting. archive.org/details/2381...

The exchanges between Adam and ChatGPT are devastating. This, in my mind, is the worst one. One of his last messages was a photo of the noose hung in his bedroom closet, asking if it was "good." ChatGPT offered a technical analysis of the set up and told him it 'could potentially suspend a human."

DEN ✈️ ORD ✈️ DCA

lunch break

DCA ✈️ DFW ✈️ DEN

Because certain AI workloads are not time-sensitive, data center flexibility is touted as a potential way to reduce strain on the grid and ensure consumers don't face increased costs. But a new study suggests data center flexibility in certain regions can actually result in higher CO2 emissions.

credulous accounts and reception to the AI Action Plan continue to shock me. they flatly do not reckon with what the Administration continually shows it wants to and will do.

credulous accounts and reception to the AI Action Plan continue to shock me. they flatly do not reckon with what the Administration continually shows it wants to and will do.

since "AI policy" is really "every issue domain policy" you can "get more alpha," as they say, by paying attention not to what the administration recommends as policy actions from the AI staffers, but what the actual wielders of power in executive agencies *do* www.techpolicy.press/trumps-ai-st...