News for hackers. Stuff that matters.
Re:OpenAI: Securing our frontier model training environment by bradleybuda on Friday October 23, 2026 @05:36PM (#50215714)
“Completely air-gapping this infrastructure—that is, disconnecting it entirely from external networks—isn’t currently feasible, since we need external connections in order to run and monitor experiments”
Re:OpenAI: Securing our frontier model training environment by angry_octet on Friday October 23, 2026 @07:23PM (#50216730)
It's incredible how much incompetence and muddled thinking is baked into their statements. Governments everywhere run fully air-gapped networks: staff have to turn up at an office, be identified to get access, and transfers in and out are carefully vetted. Communication between sites is via network encryptors with no control plane connection on the black or red side.

If they wanted to, they would.

Re:OpenAI: Securing our frontier model training environment by jerjerjer on Friday October 23, 2026 @08:17PM (#50217106)
Yes, but artifacts created in these air-gapped networks are expected to stay there. This is not true for AI lab models, who would get access to the internet once live.
Re:OpenAI: Securing our frontier model training environment by angry_octet on Friday October 23, 2026 @09:40PM (#50217624)
That is not the case, a key purpose for most intelligence functions is dissemination, not holding the secrets in the darkness. For example, satellites take pictures, mobile phones are geolocated, HUMINT sources drop dimes. Then analysis happens high side, and a version of that is automatically and certifiably downgraded to e.g. a report about troop movements, without revealing specific characteristics of the sources used.

The pipeline for model assessments could easily report summary statistics down via a trusted guard product. These are amongst the most highly validated software/hardware devices made, and provide a central point for logging and authorisation out of band.

Re:OpenAI: Securing our frontier model training environment by jerjerjer on Saturday October 24, 2026 @10:43AM (#50221735)
Sure, derived findings can be published (that's often one of the goals after all), but original artifacts are never expected to leave the premises.
Re:OpenAI: Securing our frontier model training environment by bakies on Friday October 23, 2026 @07:15PM (#50216667)
It's feasible, it'll just slow them down. Which is unacceptable to the business leaders.
Re:OpenAI: Securing our frontier model training environment by angry_octet on Friday October 23, 2026 @09:33PM (#50217581)
It wouldn't really cause more delay, there is established technology for this. They just refuse to acknowledge the scope of the problem and think they can girl boss gaslight gatekeep their way out.
Re:OpenAI: Securing our frontier model training environment by betterunix2 on Friday October 23, 2026 @06:40PM (#50216357)
I am unconvinced. Why can't they just set up workstations on the airgapped network to run and monitor their experiments?
Re:OpenAI: Securing our frontier model training environment by thomasahle on Friday October 23, 2026 @06:51PM (#50216482)
Would that require the researchers to work directly from the data centers?
Re:OpenAI: Securing our frontier model training environment by betterunix2 on Friday October 23, 2026 @06:55PM (#50216518)
Or from an office with a cable running to the data center.
Re:OpenAI: Securing our frontier model training environment by simoncion on Friday October 23, 2026 @08:45PM (#50217305)
You do the standard thing: VPN into the datacenter's management network for IPMI-provided [0] KVMs for the isolated hosts, then connect to one or more of those KVMs.

If the software under test can attack the IPMI to gain access to the management network it's connected to, and then hop out to the Internet through either the VPN server or the IPMI client software, and the problem wasn't caused by a boneheaded misconfiguration, then that's a very interesting finding that everyone will want to hear... because that's a vulnerability in a component that's expected to safely run alongside hostile workloads.

[0] I'm using the name "IPMI", but the thing goes by many other names: BMC, LOM, iDRAC, iLO, etc, etc, etc. Whatever you call it, it's a little computer built into the motherboard to provide remote access and management to the system it's built into.

Re:OpenAI: Securing our frontier model training environment by Borealid on Saturday October 24, 2026 @01:45AM (#50218617)
If you want to be even safer, you could use a physical network-connected KVM. The only interfaces those expose to the connected computer are HDMI (graphics read only) and USB peripherals (write only). Sometimes they can also physically power off the hosts to which they're connected.

It's a pretty darn slim surface for an attack. Technically doesn't meet the definition of an "air gap", but it gives you 100% of the utility of remote work with effectively 0% of the vulnerability, at the cost of a tiny amount of power and a few extra cables in the rack.

Re:OpenAI: Securing our frontier model training environment by nullc on Saturday October 24, 2026 @12:38PM (#50222904)
It's funny to juxtapose the doomerism claims of superpowerful AI when they refuse to use mechanisms already commonly used for high security systems.

You create an airgapped network and monitor the systems using that. This is not rocket surgery.