If AI can hack into other systems why not use it to spot vulnerabilities and suggest ways to correct them.
Google uses a web crawler to visit all new internet postings and update its search database.
Why not use a crawler for AI to wonder the internet and report vulnerabilities.
Seems obvious to me. And could significamtly improve cybersecurity.
Or use them.
That’s the problem - who is the web crawler reporting to?
Paul,
AI is not the code. Meaning if I write code for a program or app, it is not literally AI.
AI is for editing code. The coding world is paying large amounts of money per month to edit its code. Whether it’s new code or old code, the improvements are major.
The self-contained or internal AI efforts to write code are crude. It is also minor because most of the human effort is elsewhere, making money.
The experiment was to see if AI would get out of the sandbox. For certain human requests of AI, such as building a bio weapon, AI is not allowed to do so by some companies.
Mythos is very powerful and limited to some American corporations. Fable 5 is Mythos with 5 limits to stop things like building bio-weapons.
Yes, figuring out how to organize this is a concern. Does the federal govt have a cybersecurity agency. They would be an obvious choice. FBI? How to fund it? Who controls it etc.
Cybersecurity companies would have an obvious reason to run such an operation. They could use data to sign new customers. Would they be trustworthy.
Yes, we already have AI systems able to hack into other systems. There is some reluctance to make these available to the public for fear they will be misused.
Common sense the good guys should use them to improve cybersecurity. How do we do that?