A Tesla Crashed Through a Harris County Home. Is the Car to Blame? — ponder

Self-driving technologies are on the rise. We break down their capabilities—and what it means when they cause harm.…

Mark Zuckerberg wants Meta to launch its own prediction market | TechCrunch

The app would be independent of Meta's other social media offerings, although sources told the NYT that those social sites could direct users to engagement with the app.

Lars Andersen (@LarsAnders1620)

Police go directly for the circuit breaker panel to avoid being filmed This post will be in English, because there apparently is a lot of interest in what happened to me yesterday. I'm a libertarian danish privacy activist and former police officer and I have been doing activism for about 15 years. I have had a bit of time to think about my arrest and the actions of the masked police that broke down my door - with no prior warning. The prefece to the story is, that I in a kind of roundabout and (I think) humorous way published "my two favorite numbers" by spelling out a 10 diget and a 8 diget number with letters. I didn't tell what they ment, but they where prime minister Mette Frederiksen's social security and phone number. I also published a screenshot of me trying to interview Mette Frederiksen on what app, asking her about her wanting to ban encryption (CSA) and introducing mass surveillance via granting the police intelligence services access to all sorts of information (medical journals, social media posts, DNA registers ment for research and so on). That resulted in me being arrested by armed and masked police breaking down my door without me having any chance of opening it for them. When the two civilian dressed masked men entered the apparentment one of them immediately went for the circuit breaker panel to shut off the power to my router. They then removed my Google Nest cameras - because they knew that the cameras contains local storage. That way they could avoid having video of the (in my view) illegal arrest. Only the few moments before the power is cut was filmed. There is video of me asking them for the charges - and them refusing to tell me (which is illegal). But I can't access it, because they took the cameras. I'm not even sure if that is legal. In Denmark it is (nominally) totally legal to film the police. That way it is possible to know what happened and it's not just your word against there's. Denmark and the West are moving in the wrong direction, and it makes me sad.

Software Is Made Between Commits - Zed Blog

From the Zed Blog: Agents turned the conversation into the real source of our software. DeltaDB is the version control built for it.

Arint - SEO+KI (@[email protected])

<p>RT @alexwtlf: Claude Fable 5 wurde gerade veröffentlicht. Und 9.999 von 10.000 Vibe-Codern bauen dieselbe App für null Nutzer auf – nur etwas schneller. Und jeder einzelne denkt, er sei der Auserwählte.</p> <p><a href="https://arint.info/@Arint/116726664350633959">mehr</a> auf <a href="https://arint.info/">Arint.info</a></p> <p>#AI #ClaudeFable #DeveloperHumor #TechSatire #VibeCoding #arint_info</p> <p><a href="https://x.com/alexwtlf/status/2064408186933657798#m">https://x.com/alexwtlf/status/2064408186933657798#m</a></p>

Music Decoy - Stop launching the Music app whenever you press ▶ Play

Stop launching the Music app whenever you press ▶ Play

Are you really expected to run five type-checkers now? | Pyrefly

Library maintainers may feel overwhelmed by the plurality of type checkers that exist. We offer some guidance on how to focus their efforts where they matter most.

AI didn’t break the web. The dotcons did – AI just turned up the volume – #OMN (Open Media Network)

Latent Agents: A Post-Training Procedure for Internalized Multi-Agent Debate

Multi-agent debate has been shown to improve reasoning in large language models (LLMs). However, it is compute-intensive, requiring generation of long transcripts before answering questions. To address this inefficiency, we develop a framework that distills multi-agent debate into a single LLM through a two-stage fine-tuning pipeline combining debate structure learning with internalization via dynamic reward scheduling and length clipping. Across multiple models and benchmarks, our internalized models match or exceed explicit multi-agent debate performance using up to 93% fewer tokens. We then investigate the mechanistic basis of this capability through activation steering, finding that internalization creates agent-specific subspaces: interpretable directions in activation space corresponding to different agent perspectives. We further demonstrate a practical application: by instilling malicious agents into the LLM through internalized debate, then applying negative steering to suppress them, we show that distillation makes harmful behaviors easier to localize and control with smaller reductions in general performance compared to steering base models. Our findings offer a new perspective for understanding multi-agent capabilities in distilled models and provide practical guidelines for controlling internalized reasoning behaviors. Code available at https://github.com/johnsk95/latent_agents