Can anyone in the know ELI5 how a defence against such an attack using AI would be organized? I’m struggling to understand what actions frontier models would perceive as offensive
HN user
p1anecrazy
Hi, this is very interesting, thanks. While trying to educate myself about honeypots I came across this (https://securehoney.net/).
The aggregations of popular logins and IP locations seem interesting.
I understand the author wants to add it every time as a signature, and having claude do it doesn‘t fulfill the purpose
Is there a way for non-Japanese speakers to experience this?
Appreciate it as a regular sauna-goer. I am also struggling to wake up after sauna evenings and maybe you research explains why
Once in a while as I get sick I have to retrain myself to going to sauna (e.g. taking lowest level, even skipping the Aufguss, German infusion where temperature is raised gradually etc.)
Also IMO your body fat/water/lean muscle ratio may play a role. I once lost 5 kg due to Influenza A and all my sport achievements as well as sauna endurance were gone
One explanation I heard is “helps save water when flushing”
Really like demo cli tools description. Are they limited by the context window as well? What’s your experience with log file sizes?
Several years ago German media ran a story where they were able to connect Google search queries to members of parliament by buying big data. Examples included someone looking for data on “dementia treatment” etc. But that wasn’t real-time, as I understand.
So law enforcement was able to intercept/obtain logs of all the perpetrator’s Google Translate queries. How does that work? Assume he was not logged in. Did they just continuously obtain his IP when his device announced itself in a cell phone network and then monitored all the traffic and ordered Google to provide all the activity for that IP?
Edit: apparently his accomplice living in the U.S. was under FBI surveillance. “The use of Google Translate was not Durovic's only operational failure. He made two trips to Russia, in July and October 2024, attempting to disguise his destination by booking ostensible vacations to Turkey while continuing to Moscow on connecting flights under his real name. Each time he returned to the United States, he was questioned by FBI special agents who had access to his airline records. He flatly denied having visited Russia. The lies were transparent, and they compounded his legal exposure. The FBI, however, was content to wait — continuing to surveil him and read his virtual diary on Google Translate — before finally arresting him in March 2025.”
The report claims a single weapon of the kind was acquired in 2024 and tested on animals resulting in similar damage.
This summary differs from CBS content.
It does claim the US went to great lengths to dismiss the victims for a decade. In 2024 it obtained a single device, started testing it on animals and achieved similar effects as experienced by the victims. The victims were then invited to the White House.
100%. Original source was posted 3 times and never gained traction because it is not written for the general audience.
Well, people shouldn’t start writing word salad just because LLMs were trained on structured text.
I believe s/he means that it as if the users are leaving X for BlueSky for decentralization, but actually they are facing a similar lock-in.
What a wholesome guy. Thanks for the read
Nightmare material :(
Why do you think the ones with holes didn‘t have the same defect?
Unless it’s a repost feels it should be getting more attention
One thing is clear from this thread: you are a decent human. Thank you!
Why are you posting the same message in every thread? Is this OpenAI astroturfing?
In work chats full of AI generated slop, it stands out.
Do you mean like Teams AI autocomplete or people purposefully copying AI-generated messages into chats?
Interesting, reminds me of building bots with Lambdas and GCP Cloud functions when they were a thing.
I find sharing current location a bit of an overkill to get a user‘s timezone. What‘s your take on that?
I guess their point was that Russian military doesn‘t care if Russian intelligence reads their messages
Be the change you wish to see.
If professionals like you join European companies it will help grow their business and offer competitive salaries.
Please share the method
Just use Telegram, at least it’s not U.S. made
Can you elaborate a little, please? Also, do you use these tools for managed K8s or standard one?