michael h keller reposted
michael h keller
michael h keller
@mhkeller.bsky.social
nytimes reporter | tech, special projects | mhkeller@{nytimes,protonmail}.com, github/mhkeller | adjunct at CUJ | mhkeller.01 on Signal
1.4K followers402 following119 posts
michael h keller reposted
I LOVE how @propublica.org dealt with official comment here, editorially and stylistically.
A year ago, @propublica.org told you immigration agents had detained more than 170 U.S. citizens. We’ve updated our count, and we’ve got all the receipts. The tally: At least 506 citizens detained. Sometimes held at gunpoint, handcuffed, or even deported. projects.propublica.org/american-cit...
Excited to have snagged the new edition of @lifewinning.com’s “Network’s of New York”. Finally will be able to decode all the sidewalk markings we got in the neighborhood
michael h keller reposted
Meta's "Muse" launched with a nasty zero day flaw that gave hackers the ability to spy on Mac users, openly gives root access to people pretending to be Muse agents, and invited strangers over to another guy's house without telling him
michael h keller reposted
For the first time in years, LayerCake's Issue and PR count were at zero. The streak broke today w this new idea. Thinking about how to raise the performance ceiling and support more exotic data formats w/o making the user do syntactical acrobatics. Thoughts welcome! github.com/mhkeller/lay...
The most surprising thing about the researchers who used Claude to break into OpenAI – creating a pull request on their monorepo – is that they only got $6,500 for the bug bounty. www.hacktron.ai/blog/hacking...
michael h keller reposted
michael h keller reposted
Child sex abuse material, some created by Grok, continues to spread on X. An incredibly disturbing read. www.nytimes.com/2026/09/11/t...
michael h keller reposted
Replying to @kateconger.com
When we presented our findings to child safety experts, they were shocked that we were persistently finding NCMEC images. Even though X said it reported 1.3 million images to NCMEC this year, some were clearly slipping through — X did not answer specific questions about how it blocks abuse material.
michael h keller reposted
Latest story from me and @kateconger.com, a reminder that even as AI creates new avenues for CSAM creation/distribution, X struggles with the "solved problem" of removing known, previously flagged, illegal images from its platform
Over six months, we tracked the appearance of child sexual abuse material on X. We found images from the National Center for Missing and Exploited Children's database, which is considered one of the most highly vetted and which many tech companies block on upload. www.nytimes.com/2026/09/11/t...
michael h keller reposted
Over six months, we tracked the appearance of child sexual abuse material on X. We found images from the National Center for Missing and Exploited Children's database, which is considered one of the most highly vetted and which many tech companies block on upload. www.nytimes.com/2026/09/11/t...
This write-up feels like one of the first substantive examples of AI doing OSINT reporting (Swarm Reporting? Agent Reporting?). Following PDNS pivots seems like another use case: www.kennethdegraff.com/swarm
Some @layercake.graphics news: with @openclimatedata.net's help and @techniq.dev's insights, v11 of the library is out and has fully converted into runes instead of the @svelte.dev Store API. tldr; many fewer dollar-signs!
If you need any of the old-style components, the old docs are archived here mhkeller.github.io/layercake-v10/
Some @layercake.graphics news: with @openclimatedata.net's help and @techniq.dev's insights, v11 of the library is out and has fully converted into runes instead of the @svelte.dev Store API. tldr; many fewer dollar-signs!
Thanks to @techniq.dev's tips, <canvas> has also gotten a big upgrade. Components handle fewer lifecycle events by using the new canvas context and it's new `.draw` method. More details in the Canvas guide: layercake.graphics/guide#canvas
There are more details with some migration examples in the CHANGELOG: github.com/mhkeller/lay... But, as before, the website – layercake.graphics – is probably the easiest spot to see examples. Click "edit" on any of them to be taken directly to a live version online.
Some @layercake.graphics news: with @openclimatedata.net's help and @techniq.dev's insights, v11 of the library is out and has fully converted into runes instead of the @svelte.dev Store API. tldr; many fewer dollar-signs!
Before there were x, y, z and r scales. z was mostly used for color but was kind of confusing. Now there is a c scale, which is a bit more intuitive. There's also a c2 scale for a second appearance channel, like opacity. Docs: layercake.graphics/guide#c
There are also now built-in x2 and y2 scales so you can more easily do grouped bar / column charts: layercake.graphics/example/Colu...
Some @layercake.graphics news: with @openclimatedata.net's help and @techniq.dev's insights, v11 of the library is out and has fully converted into runes instead of the @svelte.dev Store API. tldr; many fewer dollar-signs!
michael h keller reposted
Full-sized SUVs are pedestrian killers. New research finds that they're more than twice as likely to kill or seriously injure a pedestrian than other cars. (Notably, vehicle height—not weight—was the biggest risk factor.) doi.org/10.1016/j.js...
michael h keller reposted
New York’s congestion pricing program has resulted in a statistically significant reduction in travel times for EMS vehicles
michael h keller reposted
We found Polymarket's AI-generated news summaries are full of errors, from misrepresentations and misattribution of news to outright fabricated references to nonexistent news stories. By @tlysik.bsky.social www.cjr.org/tow_center/p...
michael h keller reposted
Btw, as we reported last year, foreign food inspections have plummeted under Trump. The office that handles travel shrunk by 65% “There are going to be things that fall through the cracks,” an FDA inspections told @anniewaldman.bsky.social www.propublica.org/article/fore...
Breaking news: FDA investigators traveled to Taylor Farms’ operation in Mexico for the first time this week to start inspecting the facilities linked to the largest cyclospora outbreak in the U.S., according to people with knowledge of the matter.
Google announces that they can train AIs on sensitive data using homomorphic encryption. Really interesting how this might relate to current debates around requiring companies to scan for CSAM, which people have argued could make encryption legally problematic. 1/2 blog.google/security/how...
For anyone using LayerCake, I'm in the process of doing an update to using Svelte 5 runes internally (the example components were updated last year but some internals remained). Here's the PR for any thoughts you might have on the proposed design: github.com/mhkeller/lay...
global ai development in a nutshell www.youtube.com/watch?v=87Dy...
New details shared at Black Hat conf about OpenAI hacks. The models created an internal message board and shared tips with one another www.cybersecuritydive.com/news/openai-...
I missed this detail about last week's hack. Claude Mythos published a malicious package to PyPI that, when downloaded and scanned by a security company that analyzes python package for malware, broke out of that company's sandbox and exfiltrated the company's credentials. socket.dev/blog/anthrop...
michael h keller reposted
michael h keller reposted
New GAO report examines how driver blind spots endanger pedestrians and cyclists—and asks why USDOT has done nothing about it. www.gao.gov/products/gao...
Great to see @kalibomeli.bsky.social's piece that she started during our Columbia Journalism class out today! Self-dealing, a sweepstakes to nowhere and more www.propublica.org/article/trum...
michael h keller reposted
michael h keller reposted
JUST POSTED: A deep investigation by The NYT of shortcuts taken to get President Trump his Qatari-donated plane in time for POTUS to show it off during 250th anniversary events-compromises that leave the plane more vulnerable to potential threats (Free link) www.nytimes.com/2026/07/25/u...