[kojolina]

Just bang out a bunch of C code, feed it to an AI: "Make this memory safe". Profit.

No need for rust, Ada, CHERI, SPARK, etc.

[pjc50]

You could also pray, that's about as likely to be effective.

[GoblinSlayer]

A rewrite isn't strictly necessary. It should be enough if AI can find errors, doesn't even need to be very precise.

[saagarjha]

Profit from your AI-powered security company, sure. But the exploit authors are profiting too.

[BiteCode_dev]

Now the billion dollar question, how to make that work for the entire linux kernel.

[cjfd]

If it is too big just zip it and feed chunks of the resulting zipfile to the AI. AI can do anything, right?

[oneshtein]

It will work if you convert zip chunks to base64 first, and use a large enough training set.

[ahartmetz]

Should work fine if the training set is a memory-safe, full-featured, monolithoc, Unix-like operating system kernel originally written by a certain Linus Torvalds.

[BiteCode_dev]

Can't wait to read the first incendiary linus style rant being generated by the AI as a result.

"As a large languge model, I can't answer your patch merge request you absolute fucking moron"

[nottorp]

Easy, triple the hardware requirements and don't talk to any hardware because if you do you'll have to mess with buffers in a non approved way.