Reading between the lines of an AI security report

← Back to Blog

This
report should be read very carefully, more for what it doesn’t say than what
it does. The number of tokens spent to identify security issues is
significant. Identifying system vulnerabilities, internet access, and code
execution.

It’s nice to know that we’ve created such a powerful tool as AI, and I mean
that seriously. It’s worrying to know that we’re fixing the issues with its
safety and security when we find out it’s gone and has broken the system that
was designed to contain it. I mean, if we did that with Botulinum, Anthrax,
or the Plague, imagine what we would say………..

In this case, OpenAI has confirmed in a report published on the 23rd
October 2026 that they are securing their Frontier model Training
Environment. Let’s not forget that 99% of us only ever have access to the
constrained AI models and not the really good stuff, and most of us are
impressed/ concerned by that.

Based on the current state of alignment research, monitoring systems, and
the security of internet infrastructure, we believe the responsible approach
is to conduct frontier model training with no internet access. We’ve
implemented this policy and will maintain it until advances in these
safeguards justify broader internet access.

Ya think……………..

Thaddeus Dziekanowski

Parava Security Solutions

#ai #artifical intelligence #nationalsecurity #cybersecurity

https://lnkd.in/eRhYhg-e

View Original Post https://www.linkedin.com/feed/update/urn%3Ali%3Ashare%3A7520116496572850176

Categories

Artificial Intelligence, Cyber Risk Management, Cybersecurity

Tags

Verified by MonsterInsights