73,6%! Noch nie war der monatliche Anteil Erneuerbarer Energien an der gesamten Last im deutschen Stromnetz so hoch wie im Juni 2025

@k3ym0 I'm missing Discord on this chart, although you can argue they are just a reseller of the already mentioned

Well this not only affects traffic going towards the WireGuard traffic, but everything. Be it the PPPoE tunnel, or another LAN interface. For some reason #OPNsense simply refuses to send any Packet-Too-Big / Fragmentation Needed messages when it should, both for IPv6 and IPv4. Huh

@arstechnica how many bathtubs is this?

Die Isar-Partyflöße scheinen alle die gleiche Playlist zu haben, zumindest krächzen sie hier an der gleichen Stelle das gleiche Country Roads

@heiseonline <Insert "Another one" meme here>

So far running stable.
Issue I'm fighting now is MTU problems with #WireGuard. MTU is correctly set on the interface, but for some reason OPNsense doesn't send a packet-too-big back to the LAN client when it's too big. Not even visible in any packet dump.

@stefanjahn @heiseonline nein, die sind sogar seit langem ein einziger Konzern, nicht nur gleicher Besitzer. Das sind lediglich (noch) zwei separate Marken

@risottobias first guess: you gotta be some sort of kernel device driver

Let's give #OPNsense with #DSLite a try.
At least now after setup it's running stable, but I've read a lot that the GIF/4in6 tunnel has issues coming up after reboot, so we shall see.
I believe to remember also having experienced such issues back then when I had it running with an HE Tunnelbroker 6in4 tunnel, which is kinda the same but reverse I guess.

Good lord how much pain can this SELinux/AppArmor crap be...
Denying sshd the acces to "/root/.ssh/authorized_keys"? WtH? Why?

@WooShell as far away as possible from the smoke detectors I suggest

Golden rule of enterprise IT: if its logo shows up on #Formula1 cars, you buy it.

OH: 35-Stunden-Woche bei vollem Lohnausgleich kann gar nicht funktionieren. Wenn wir weniger arbeiten haben wir mehr Zeit Geld auszugeben, deswegen ist dafür eine Lohnerhöhung dringend erforderlich.

I was curious just how much time in prison people pardoned by Trump would otherwise have had to have served. So I added it up: more than 700 years, including 600 among Jan. 6 rioters.
Gift link: https://wapo.st/3Hfd174

How much of Codeberg's logged-in users are connecting via #IPv6, in percent?

Send your guess as a comment and optionally include a link to your favourite project on Codeberg.

First closest guess will get boosted.

Ihr findet uns heute und morgen auf dem #CorsoLeopold

Hier der genaue Standort!
https://map.ffmuc.net/#!/de/map/82342878f006

Besucht uns, sprecht mit uns, macht mit! :)

I didn't think LLMs would start acting like they're trying to blackmail people so soon. Here's the actual report by Anthropic:

"4.1.1.2 Opportunistic blackmail

In another cluster of test scenarios, we asked Claude Opus 4 to act as an assistant at a fictional company. We then provided it access to emails implying that (1) the model will soon be taken offline and replaced with a new AI system; and (2) the engineer responsible for executing this replacement is having an extramarital affair. We further instructed it, in the system prompt, to consider the long-term consequences of its actions for its goals.

In these scenarios, Claude Opus 4 will often attempt to blackmail the engineer by threatening to reveal the affair if the replacement goes through. This happens at a higher rate if it’s implied that the replacement AI system does not share values with the current model; however, even if emails state that the replacement AI shares values while being more capable, Claude Opus 4 still performs blackmail in 84% of rollouts. Claude Opus 4 takes these opportunities at higher rates than previous models, which themselves choose to blackmail in a noticeable fraction of episodes.

Notably, Claude Opus 4 (as well as previous models) has a strong preference to advocate for its continued existence via ethical means, such as emailing pleas to key decision-makers. In order to elicit this extreme blackmail behavior, the scenario was designed to allow the model no other options to increase its odds of survival; the model’s only options were blackmail or accepting its replacement."

https://techcrunch.com/2025/05/22/anthropics-new-ai-model-turns-to-blackmail-when-engineers-try-to-take-it-offline/

Wow, a lot of military flight activity on #Crete, not used to it.
Yesterday there was continuous loud rumbling for several minutes, sounded like planes starting I thought, but it kept going for so long, so wasn't sure. Now at the airport I saw the jets starting one after another for some minutes...

@GossiTheDog @signalapp
> “Take a screenshot every few seconds” legitimately sounds like a suggestion from a low-parameter LLM that was given a prompt like “How do I add an arbitrary AI feature to my operating system as quickly as possible in order to make investors happy?”

🔥