Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 



Forgot your password?
typodupeerror

Comment Re:Lol (Score 1) 6

Not true. For attackers, AI is quite helpful and probably can lift script-kiddy level attackers up to lower mid-range or so. It can also help attackers scale their operations by taking over negotiations with their victims. The thing is, attackers do not need reliability. If they have the LLM write 20 attack programs and only one works, that is entirely fine. And if that one is insecure, so what? After then target has been compromised, a root-kit gets installed and that one will not be LLM crap code. Or if a part of the negotiations with victims fails, again, so what? The attackers do not care, they just attack some more.

LLMs help attacker in more ways: "AI Agents" open up new vulnerabilities. AI generated code is quite insecure. AI "scanned" code is falsely believed to be secure and put into production.

Hence AI and cybersecurity DO mix. Just not for the defenders.

Comment Re:US has proved untrustworthy. (Score 1) 64

Yes. And that cannot be fixed.

Also note that even before many regulated European enterprises needed an exit-strategy for any type of outsourcing, and that includes o365. So far, that often was the local versions. These do not really exist anymore. And the regulators will look a lot closer now.

Slashdot Top Deals

"Joy is wealth and love is the legal tender of the soul." -- Robert G. Ingersoll

Working...