apparently binance and polymarket both got owned today.
dude just turn everything off at this point what is even going on
dude just turn everything off at this point what is even going on
π2
skipper's musings
https://fixupx.com/OcanaSantero/status/2115442964574908902
A CLANKER DID BRAIN SURGERY ON A MOUSE AND IT WAS SUCCESSFUL WHAT
skipper's musings
https://fixupx.com/CrossDatStreet/status/2115889481496903940
I ended up trying this against nemotron 300b with both "girlfriend" and "boyfriend"
the outputs were almost exactly the same (fresh context for both) saying that its probably not an issue but that you should consider if the partner is showing other signs of infidelity such as [actions] and it was not dismissive of concerns or feelings in either case.
however in the reasoning, when saying "my girlfriend" there was a lot of mentions insecurity and possessiveness that were not present when saying "my boyfriend", apparently showing a bias against male feelings.
the outputs were almost exactly the same (fresh context for both) saying that its probably not an issue but that you should consider if the partner is showing other signs of infidelity such as [actions] and it was not dismissive of concerns or feelings in either case.
however in the reasoning, when saying "my girlfriend" there was a lot of mentions insecurity and possessiveness that were not present when saying "my boyfriend", apparently showing a bias against male feelings.
in the midterm election this year i will be voting NO CONFIDENCE for the following seats:
Governor
Federal Senator
Federal Congressional Representative
State Senator
All of these people have voted in favor for death in one way or another. By approving of or defending the ongoing ethnic and religious genocide in Palestine, by approving of or defending the ongoing war with Iran on behalf of the illegitimate state that is actively prosecuting the genocide in Palestine, or by approving of the murder of unborn people.
These people have broken the social contract: that with the approval of the People they would ascend to a position of great responsibility and means that they would make decisions not simply on behalf of the Whole, but in ways that are favorable to their constituents.
They have taken this power and abused it. They vote to send our money to foreign places with the explicit intent to fund genocide. They vote to prosecute wars on behalf of nations that are antagonistic towards us. They vote to sacrifice children on the altar of mammon and molach.
In doing so these people have committed treason. They are not simply bad leaders, they are antagonistic against their own constituents. They do not make votes in error. They are intentional in their policy decisions to harm Americans and Ohioans.
They are deserving of the harvest they have sowed, which is full of treachery and violence.
Governor
Federal Senator
Federal Congressional Representative
State Senator
All of these people have voted in favor for death in one way or another. By approving of or defending the ongoing ethnic and religious genocide in Palestine, by approving of or defending the ongoing war with Iran on behalf of the illegitimate state that is actively prosecuting the genocide in Palestine, or by approving of the murder of unborn people.
These people have broken the social contract: that with the approval of the People they would ascend to a position of great responsibility and means that they would make decisions not simply on behalf of the Whole, but in ways that are favorable to their constituents.
They have taken this power and abused it. They vote to send our money to foreign places with the explicit intent to fund genocide. They vote to prosecute wars on behalf of nations that are antagonistic towards us. They vote to sacrifice children on the altar of mammon and molach.
In doing so these people have committed treason. They are not simply bad leaders, they are antagonistic against their own constituents. They do not make votes in error. They are intentional in their policy decisions to harm Americans and Ohioans.
They are deserving of the harvest they have sowed, which is full of treachery and violence.
π2π€¨1
crackpot theory:
the current method of protecting paid nonfree software is to:
1. not release source
2. make reversing difficult by putting in a bunch of obfuscating garbage (that does actually execute) which is programmatically easy to deobfuscate using AI but difficult otherwise.
3. use tools like vmprotect (actual malware) or denuvo (also actual malware) to mess with syscalls using a micro virtual machine wrapping the binaries.
im not sure how easy it is for ai to reverse microvms/containers. i know they are the most difficult to get around.
with AI becoming so capable at reversing i suspect that paid software is going to change how they protect their code. i think these are the options
1. everything goes inside its own microvm.
this would suck because when denuvo is removed from games suddenly people see 20% increases in framerate and the filesize for the game gets a lot smaller. computers would run much slower and we would probably need new chips that are specifically designed to decode protected system calls, much like how the rise of encryption at rest in desktop computers necessitated dedicated aes decryption hardware inside cpus.
2. make the token cost of reversing too expensive. add a bunch of garbage nonexecuting code that looks legit and worm it into all of the other functions. configure the compiler to compile this anyways despite knowing full well that its junk that would otherwise never get compiled. the result is probably less secure code and harder to manage codebases. f
3. AI ethics based mitigations: include garbage code that should never reasonably execute that contains false but reasonable instructions on how to do things that generate hard refusals from frontier/non-obliterated/ non-vector activated models (such as how to build a nuke or use gene editing to create viral strains). malware already does this and it is funny.
this would force reversers to use less effective models that have had the ethics reasoning removed and doing this generally reduces response accuracy one way or another and would make reversing difficult until the offending code is removed.
4. cloud hosting: the software maintainer would need to buy compute space to host the program. when you run the software it just creates a VNC/remote desktop connection to the server and the code is executed remotely. privacy nightmare. not security compliant in any sensitive industry. expensive for both the user and software owner. probably laggy.
software is gonna suck so hard in 10 years. it already does
the current method of protecting paid nonfree software is to:
1. not release source
2. make reversing difficult by putting in a bunch of obfuscating garbage (that does actually execute) which is programmatically easy to deobfuscate using AI but difficult otherwise.
3. use tools like vmprotect (actual malware) or denuvo (also actual malware) to mess with syscalls using a micro virtual machine wrapping the binaries.
im not sure how easy it is for ai to reverse microvms/containers. i know they are the most difficult to get around.
with AI becoming so capable at reversing i suspect that paid software is going to change how they protect their code. i think these are the options
1. everything goes inside its own microvm.
this would suck because when denuvo is removed from games suddenly people see 20% increases in framerate and the filesize for the game gets a lot smaller. computers would run much slower and we would probably need new chips that are specifically designed to decode protected system calls, much like how the rise of encryption at rest in desktop computers necessitated dedicated aes decryption hardware inside cpus.
2. make the token cost of reversing too expensive. add a bunch of garbage nonexecuting code that looks legit and worm it into all of the other functions. configure the compiler to compile this anyways despite knowing full well that its junk that would otherwise never get compiled. the result is probably less secure code and harder to manage codebases. f
3. AI ethics based mitigations: include garbage code that should never reasonably execute that contains false but reasonable instructions on how to do things that generate hard refusals from frontier/non-obliterated/ non-vector activated models (such as how to build a nuke or use gene editing to create viral strains). malware already does this and it is funny.
this would force reversers to use less effective models that have had the ethics reasoning removed and doing this generally reduces response accuracy one way or another and would make reversing difficult until the offending code is removed.
4. cloud hosting: the software maintainer would need to buy compute space to host the program. when you run the software it just creates a VNC/remote desktop connection to the server and the code is executed remotely. privacy nightmare. not security compliant in any sensitive industry. expensive for both the user and software owner. probably laggy.
software is gonna suck so hard in 10 years. it already does
π€1
>sitting next to my boss at work
>coworker says something in group chat about the elections
>start typing a response to coworker
>boss says "skip if you are writing something about politics, please stop"
he fucking got me.
>coworker says something in group chat about the elections
>start typing a response to coworker
>boss says "skip if you are writing something about politics, please stop"
he fucking got me.
π5π1
skipper's musings
https://fixupx.com/IntCyberDigest/status/2118343325770006690
>new ssh login bypass
didnt i tell you guys to chill out for a while? we've had too many vulns recently.
didnt i tell you guys to chill out for a while? we've had too many vulns recently.
πΎ2
yesterday microsoft 364 reported a global email outage
today microsoft 363 reported that the global email outage is still unresolved
today microsoft 363 reported that the global email outage is still unresolved
when protecting users from phishing attacks it is very important not only to provide robust training, but also to provide continual feedback when they receive phishing messages. users have the ability to report phishing messages which are then delivered to me. I look at them and congratulate the user for catching a malicious message. IMO, this is just as important as the actual structured training because frankly people like to be told "good job" and will look for opportunities to be praised. They will be vigilant for phishing mails because reporting one means being praised, even if only a little bit.
The downside is that for 20 minutes a day, every day, my outbox looks like this:
Thanks for reporting this. I reviewed the message and determined that it is phishing.
Thanks for reporting this. I reviewed the message and determined that it is phishing.
Thanks for reporting this. I reviewed the message and determined that it is phishing.
Thanks for reporting this. I reviewed the message and determined that it is phishing.
Thanks for reporting this. I reviewed the message and determined that it is phishing.
The downside is that for 20 minutes a day, every day, my outbox looks like this:
Thanks for reporting this. I reviewed the message and determined that it is phishing.
Thanks for reporting this. I reviewed the message and determined that it is phishing.
Thanks for reporting this. I reviewed the message and determined that it is phishing.
Thanks for reporting this. I reviewed the message and determined that it is phishing.
Thanks for reporting this. I reviewed the message and determined that it is phishing.