Fun story - About 10 years ago I was reversing a “new” malware that was monitoring all keystrokes on users computer but didn’t find it doing anything malicious. Turns out it was a keylogger installed by the banks that was required for customers to login to the bank account and for long time SK banks would only work on IE6

I've been trying to get in touch with South Korea's government IT. Have got really severe vulnerabilities that could literally compromise national security. No response so far but maybe been emailing the wrong places. If anyone could point me a direction, please shoot me an email.

They didn't have this problem when they forced everyone to use Internet Explorer + ActiveX...

I’m curious to see their evidence on how they knew it was AI agents and which agents were used.

Doesn't SK have famously, laughably terrible internet security by law?

SK is such a crazy arc. Somewhat recently got over famine and authoritarianism, then five families took over the whole country, while growing the economy to crazy levels per capita, in the fastest time ever.

They took over global pop music, while also entering demographic collapse faster than anyone.

In 2024 they had their own Jan 6th type event, but democracy appears to have won? Yet last month, they sent 30% of Russia's diesel imports via their ports, while very loudly complaining about the public announcement of two captured NK soldiers in Ukraine.

I am very ignorant, and I genuinely have no idea what is happening in SK. But last month, they were helping fund the NK + Russian war machine? What a trip.

I recently read reports about Korean banks that were not affected by the latest hacking incident.. They restrict work access to designated tablets only, and they don’t give loan recruiters access to the internal network or even a separate work system. When a recruiter hands a case over to a branch, the rest of the process is handled internally.

Looking at this, it seems that rather than making things easier because of AI, it actually requires more bureaucratic handling. The surface that made people comfortable is, paradoxically, becoming AI’s attack surface.

Here’s my bet: open weights AI will become powerful enough that it will be used by adversaries to create havoc.

This is the main risk that the labs OpenAI and Anthropic have called out for and asked for slowing the frontier.

Almost all people thought that this was fear mongering, hype marketing and regulation capture.

The same people will blame OpenAI and Anthropic for creating this.

Waiting for the day AI agents break into Nasdaq to achieve a goal.

We dont need banks. Interest bearing CBDC wallets is all most people need. Richie rich already hire people to keep tabs on their assets.