[Discuss] With AI, 2, 000 Vulnerabilities per Linux kernel release

[richard.pieri at gmail.com (Rich Pieri)]

On Mon, 14 Sep 2026 10:02:13 -0700
Kent Borg <kentborg at borg.org> wrote:

> Do you chock the wheels of your Tesla when not in use?

"Better" still... okay, a brief digression.

This ties into the lies the AI tech bros keep spouting about not
understanding why the Hugging Face attack happened. They know *exactly*
why it happened: neural networks are not programmed. They are trained,
and this model was trained to escape the jail by every means possible.

Then it was instructed to escape the jail.

So that's what it did. Mission accomplished.

Now imagine that instead of attacking Hugging Face, an offensive model
like this commandeers every connected vehicle within a 25 mile radius
to ram the gates and walls of some target building to achieve the
specified goal. Do I think it will happen? Yes. I think it will happen,
or something like it will happen, because the tech bros don't care
about anyone but themselves. They don't want Culture Minds. They want
Skynet and the Torment Nexus

-- 
\m/ (--) \m/