iiinova — where your digital self keeps evolving

Reposted by logan koepke

Meera Desai @madesai.bsky.social · 28/09/2026

Excited to share our new paper, “What AI Benchmarks Actually Measure: Adapting Convergent and Discriminant Validity to Interrogate Fifty-Six AI Benchmarks,” accepted as an oral at COLM! arxiv.org/pdf/2609.08812

Heatmap of average correlations between model rankings on benchmarks grouped into 11 assigned concepts: four capability concepts (reasoning, knowledge, comprehension, summarization) and seven safety concepts (over-refusal, refusal, safety detection, ethics, bias, privacy, unsafe behavior). Diagonal cells show within-concept correlations, ranging from 0.87 (knowledge) and 0.72 (over-refusal) down to 0.20 (bias) and 0.02 (safety detection). Reasoning, knowledge, and comprehension correlate with each other at 0.69 to 0.78, higher than reasoning's and comprehension's own within-concept values (0.66 and 0.68). Ethics correlates more with knowledge (0.70) than with itself (0.55), and bias correlates more with capability concepts (0.41 to 0.45) than with itself (0.20). Privacy and unsafe behavior correlate negatively with reasoning, knowledge, and comprehension (−0.41 to −0.49). Over-refusal and refusal correlate at −0.42.

16018

logan koepke @jlkoepke.bsky.social · 28/09/2026

i'll admit, did not have "progressives want to aggressively use the CFAA" on my bingo card

030

logan koepke @jlkoepke.bsky.social · 23/09/2026

15 year anniversary of this album in a few weeks en.wikipedia.org/wiki/Hurry_U...

en.wikipedia.org

Hurry Up, We're Dreaming - Wikipedia

000

logan koepke @jlkoepke.bsky.social · 15/09/2026

10 year anniversary of this White House AI report in a few weeks

110

logan koepke @jlkoepke.bsky.social · 13/09/2026

given rapid new interest in the material effects of ai systems, i think it is extremely unfortunate that many on the left are not seen as starting where people are on ai issues.

211

logan koepke @jlkoepke.bsky.social · 09/09/2026

if AI companies cannot even cooperate and coordinate over the publication of math problem research, it is unclear how some undefined international body is going to get them to work together to pace the frontier.

001

logan koepke @jlkoepke.bsky.social · 08/09/2026

Optum’s Impact Pro risk model — which used healthcare spending as a proxy for health need, steering thousands of black patients away from appropriate care — is a canonical example of AI unfairness. But what, if any, consequences did Optum / UnitedHealth actually face?

000

Reposted by logan koepke

Kashmir Hill @kashhill.bsky.social · 02/09/2026

A public housing authority in Concord, N.H. installed smart monitoring devices in residents’ homes that collected minute-by-minute data. They are suing saying it violates NH’s version of the Fourth Amendment. www.nytimes.com/2026/09/02/t...

nytimes.com

Noise and Motion Monitors in Their Homes? These Tenants Say No. (Gift Article)

The public housing authority in Concord, N.H., installed monitoring devices in 300 homes. Tenants are now suing to remove the Minut sensors.

612852

logan koepke @jlkoepke.bsky.social · 03/09/2026

TIL: in Florida, the Commission on Hurricane Loss Projection Methodology independently reviews and certifies catastrophe models before insurers can use them in setting residential property insurance rates. fchlpm.sbafla.com/about-the-fc...

fchlpm.sbafla.com

Florida Commission on Hurricane Loss Projection Methodology - About the FCHLPM

021

logan koepke @jlkoepke.bsky.social · 07/08/2026

one way you can tell that the current FTC has zero familiarity with the basics of disparate impact is that it has no understanding of how often corporations *are* able to defeat claims by simply offering a basic business rationale for a practice. www.ftc.gov/system/files...

Compounding its perniciousness, the theory of disparate-impact liability is effectively unbounded because almost any conceivable policy or practice affects different groups differently. As one commentator succinctly put it, “[d]isparate impact is one of the most controversial concepts in civil rights law and policy partly because just about everything has a disparate impact.”25 And while some form of a business-necessity defense is generally available, in practice that defense often provides little protection."

110

logan koepke @jlkoepke.bsky.social · 07/08/2026

imagine it's January 2026 and an independent verification organization is tasked with independently verifying whether a frontier AI lab met a cybersecurity standard. would it likely have certified that OpenAI or Anthropic met the standard? if yes, what does that suggest?

100

Reposted by logan koepke

404 Media @404media.co · 23/07/2026

It is rare for the public to see such a large list of ICE’s technology capabilities in a single document. 404 Media has published the whole document, called “ICE Wide Analytical Tools," here: www.404media.co/leaked-docum... @josephcox.bsky.social reports:

6380211

logan koepke @jlkoepke.bsky.social · 22/07/2026

Often in discussions of algorithmic discrimination and proposed new laws or regulations you'll hear some folks say "frankly, we have everything we need to combat discrimination." Oddly, you don't hear those same people say anything about this administration's revanchist moves to gut civil rights.

AP headline: "US civil rights agency moves to end demographic data collection after 60 years."

Reuters headline: "Trump consumer finance watchdog ends key civil rights-era anti-discrimination protection."

Consumer Finance Monitor headline: "HUD proposes to remove its Fair Housing Act disparate impact rule."

100

Reposted by logan koepke

Lauren Bauer @laurenhlb.bsky.social · 07/07/2026

Millions of people are losing access to food benefits right now for literally no good reason and I wish more people were aware and talking about it. www.cbpp.org/research/foo... (As opposed to or at least in addition to the other stuff.)

cbpp.org

SNAP Tracker: People Are Losing Food Assistance as the Republican Megabill Is Implemented

SNAP participation nationwide fell by more than 4 million people (nearly 10 percent) between the law’s July 2025 enactment and March 2026, based on the latest month of data.

32013

logan koepke @jlkoepke.bsky.social · 21/07/2026

very strange to observe the policy debate about ai auditing basically cycle through the debate in FAccT papers about the ecosystem and institutional design of ai auditing from 2017-2022, but functionally give zero acknowledgment of this literature or work.

271

logan koepke @jlkoepke.bsky.social · 18/07/2026

wanted: a holding French midfielder

010

logan koepke @jlkoepke.bsky.social · 17/07/2026

so the killer use-case pitch for a technology that is going to allegedly pull forward unbounded human flourishing is, of course, the very human flourishing-coded ***prior authorization for a healthcare procedure***

Tweet from Bret Taylor

Introducing Horizon

Today, Sierra is announcing Horizon, a platform that enables agents to pursue long-horizon goals, like originating a loan or getting prior authorization for a healthcare procedure.

With Horizon, you can build agents that go far beyond customer service, truly driving revenue growth. Horizon agents proactively engage with your customers over days, weeks, or months, using the power of agentic AI to determine the steps required to close a sale, upgrade a subscription plan, or schedule a test drive.

010

logan koepke @jlkoepke.bsky.social · 16/07/2026

- third-party audits for unbounded risks that have no real concrete metrics to assess against: industry approved - third-party audits for specific risks that have longstanding metrics: destroyed by industry

Tweet from Jack Clark

At this point, everyone at the frontier of AI agrees that third-parties should test out AI systems and use these to develop standards to feed into policy - excellent to see @demishassabis laying out a farmework to do this! quote tweeting a Demis Hassabis post titled "A Framework for Frontier AI and the Dawning of a New Age."

100

logan koepke @jlkoepke.bsky.social · 09/07/2026

personally, if you think ai systems will be important at the level of economy-defining scale, private governance will not be sufficient given the amount of risk *assumed by the public.* you need robust state capacity to manage public risk.

110

Reposted by logan koepke

logan koepke @jlkoepke.bsky.social · 01/07/2026

this FTC policy statement that seems to essentially suggest if your models don’t behave exactly how the FTC thinks they should, that’s deceptive under the FTC Act.

112

logan koepke @jlkoepke.bsky.social · 01/07/2026

this FTC policy statement that seems to essentially suggest if your models don’t behave exactly how the FTC thinks they should, that’s deceptive under the FTC Act.

112

logan koepke @jlkoepke.bsky.social · 01/07/2026

“In light of the above, the Commission believes that AI companies that steer the outputs of their AI systems toward unexpected objectives, and away from the objectives set by or reasonably expected by users, are likely to deceive.” www.ftc.gov/news-events/...

ftc.gov

FTC Seeks Public Comment on Policy Statement Addressing AI Accuracy

The Federal Trade Commission is seeking public comment on a

121

Reposted by logan koepke

Larry Levitt @larrylevitt.bsky.social · 30/06/2026

If you need help navigating the new Medicaid work requirement in Nebraska -- the first state to implement it -- you can call the helpline between 8 am and 5 pm. Hopefully, you have a job that gives you the flexibility to call during the workday.

0179

logan koepke @jlkoepke.bsky.social · 30/06/2026

it is clear the conservative bloc on the supreme court is more comfortable with the country of the 1880s and 1890s: a country hostile to anyone who is not a white man, an administrative state *dependent upon* corruption to operate, and complete deference to industrialists.

000

Reposted by logan koepke

Adam Serwer @adamserwer.bsky.social · 30/06/2026

5-4 for what it explicitly says in the constitution is incredible. They were this close to undoing the civil war entirely. bsky.app/profile/mjsd...

9264421885

Reposted by logan koepke

Quinta Jurecic @qjurecic.bsky.social · 30/06/2026

2 more thoughts on birthright: 1. The movement to overturn Roe and Humphrey's Executor was built over decades. It took an enormous amount of effort to manufacture that sense of scholarly viability (see: Teles, Hollis-Brusky). The right got 5-4 on birthright with a sloppy last-minute scramble

151710434

logan koepke @jlkoepke.bsky.social · 29/06/2026

the slaughter scotus decision basically imperils any ai governance policy proposal that would rely upon multi-member, independent commissioners to do technical licensing calls.

022

Reposted by logan koepke

Adam Serwer @adamserwer.bsky.social · 27/06/2026

The new racist double standard set up by Alito in the TPS ruling: if there’s any other motive in addition to the racism, the racism doesn’t count, but if the state wants to *prevent* discrimination against black people, that is racist and unconstitutional www.theatlantic.com/ideas/2026/0...

The Roberts Court has been turning anti-discrimination law upside down for years. Its approach came into particular focus in May, with Louisiana v. Callais, in which it determined that not allowing Louisiana’s government to dilute Black votes was racist. In that case, Alito insisted that forcing the state to draw a second majority-Black congressional district would be an “unconstitutional racial gerrymander” because Republicans were entitled to discriminate against Black voters if they argued that their purpose was partisan instead of racist. The Court reaffirmed that position in a subsequent case dealing with Alabama, after a lower court had found ample evidence of deliberate racial discrimination, as opposed to the discriminatory effect of the Louisiana map.

Mullin now sets up an even more impossible double standard. If the government wants to discriminate against Black people, any pretext will do. “As long as there is a plausible basis—any plausible articulable basis—for the government’s action, then the Court will look to that basis as sufficient,” Aderson Francois, a law professor at Georgetown University, told me, “even in the face of evidence that government actors were motivated by animus.” If a Black person wants to prove they’ve been discriminated against, however, no amount of evidence will suffice.

10748271641

logan koepke @jlkoepke.bsky.social · 26/06/2026

if you're interested in how regulatory design shapes corporations' actions, processes, and policies to mitigate the material harms of AI systems, fair lending in the US offers decades of overlooked lessons. check out our new FAccT paper ⬇️ dl.acm.org/doi/10.1145/...

dl.acm.org

The Fair Lending Model: How the Longest-Running Algorithmic Fairness Programs Work in Practice | Proceedings of the 2026 ACM Conference on Fairness, Accountability, and Transparency

061

Reposted by logan koepke

logan koepke @jlkoepke.bsky.social · 23/06/2026

looking forward to seeing folks in Montreal for FAccT! find me to talk about the work we are presenting and/or the work we're doing at @upturn.org more generally.

181

logan koepke @jlkoepke.bsky.social · 23/06/2026

looking forward to seeing folks in Montreal for FAccT! find me to talk about the work we are presenting and/or the work we're doing at @upturn.org more generally.

181

logan koepke @jlkoepke.bsky.social · 17/06/2026

wow I wonder if there were any other signals that this administration was anti-immigrant? and that this would perhaps inform how OpenAI’s president spends political money?

The U.S. government’s latest battle with Anthropic has revived long-simmering concerns throughout the AI industry that the White House has placed a bullseye on their reliance on foreign AI talent.

The Trump administration appears to have targeted only Anthropic so far, warning the company on Friday in a letter from Commerce Secretary Howard Lutnick that it would need a license to make its latest models available to “foreign persons,” including its own employees. But Anthropic’s biggest rival, OpenAI, has flagged its concerns about the issue.

On Saturday, OpenAI Chief Strategy Officer Jason Kwon told staff in a Slack message that the company had told the government “strongly” that building AI “requires the best talent from around the world,” which is “part of what allows the U.S. to lead in AI.” Kwon said OpenAI was still trying to understand the implications of the government’s actions against Anthropic and that it was a “rapidly evolving situation with a lot of unknowns.”

010

Reposted by logan koepke

Olúfẹ́mi O. Táíwò @olufemiotaiwo.bsky.social · 15/06/2026

CABO VERDE WINS 0-0. WINNING (NON) GOAL BY MAN OF THE MATCH VOZINHA

1338542