Democratic Sen. Mark Warner of Virginia spoke a few weeks ago about Anthropic’s “Mythos model” of artificial intelligence. Warner said.“This tool penetrated almost all of our classified systems, not in weeks but in hours.”
Hmmm.
An artificial intelligence system can penetrate almost every classified system in the United States within hours.
This seems quite dangerous.
Am I allowed to ask a follow-up question?
How does the Mythos model perform when asked to focus on classified North Korean, Iranian, Russian, or Chinese systems?
From the sounds of things, you could give the Mythos model a few days of scares and all of North Korea’s nuclear weapons would be disabled. The Middle East would not have to worry about Iranian ballistic missiles, because none of them would fly.
RIGHT?
Did I miss something?
I mean, if the problem is that our computer doesn’t speak Persian, surely that could be fixed.
Maybe the computer doesn’t speak Persian anymore.
But adjust the large language model slightly, and the computer actually speaks Persian… now.
Are the systems used by the US government to protect its classified documents so inferior to the systems used by other countries that Mythos can hack our systems, but not theirs?
If so, shame on us.
But you just have to harness the AI somehow and make it work.
Think about it: we don’t have to worry about Russia’s nuclear arsenal because we’ve set the AI to disable it. We don’t have to worry about the Chinese stealing our intellectual property because all the stealing software has been disabled. In fact, maybe we could start stealing Chinese intellectual property to build electric vehicles, solar panels, robots, etc.
To pull! Maybe AI could figure out how to make the Lincoln Memorial’s reflecting pool blue again.
I’m just a stupid keyboard clown, typing in the cheap seats here.
But what if I had this thought: If AI can hack our computers, can’t it hack theirs? —Have at least some people in the American government not had the same thought?
Did these people put their ideas into practice?
Perhaps the United States becomes hegemonic again. All other countries have been infiltrated; their systems will not work; and they don’t know it yet.
Or maybe Senator Mark Warner misstated things slightly.
Damn if I know.
Ask ChatGPT.
Mark Herrmann spent 17 years as an associate at a leading international law firm, then oversaw litigation, compliance and employment matters at a large international corporation. He is the author of The Curmudgeon’s Guide to Practicing Law And Drug and device product liability litigation strategy (affiliate links). You can reach him by email at (email protected).
