https://www.techspot.com/news/112677-nsa-using-anthropic-claude-mythos-offensive-cyber-ops.html

@kimmonismus This are spaghetti relationships topped with a sauce of their own making.

@kimmonismus just matter of time before anthropics red lines shift, government $ hard to resist

@kimmonismus Lol wadahelly is even happening

@kimmonismus I really hope Anthropic manages to keep its stance on what it will allow its models to be used for.

@PaulGugAI yeah, makes sense

@kimmonismus NSA is probably under DoD, that’s why DoW supply chain risk need not apply.

Feels less like a contradiction and more like classic US gov fragmentation: procurement risk frameworks on one track, intel ops pragmatism on another. “Blacklisted” in one lane, indispensable in another, with no real coordination layer.

What I’m wondering: is the real story here the specific Mythos use case, or the governance gap between agencies that lets this kind of split-brain persist?

@kimmonismus strange time for Anthropic now

@kimmonismus Claude Mythos is dangerous if used in war.

@kimmonismus might be wrong but the 'contradiction' framing assumes pentagon procurement and NSA ops actually coordinate vendor decisions do they? feels like those are just two separate bureaucratic tracks doing their own thing

@kimmonismus

@kimmonismus the pentagon and nsa really need to compare notes at lunch sometime

one hand doesnt know what the other is doing type beat

@kimmonismus the DoW classification cuts federal procurement but doesn't touch intelligence budgets, which is the whole point

NSA buys what works, not what passes interagency compliance. they ran Kaspersky internally years after the federal ban too

@kimmonismus Left hand blacklists Anthropic, right hand deploys it for offensive ops. Bureaucratic consistency at its finest.

@PawelJLisowski claude is too good, literally

@kimmonismus so the pentagon says no trust, but the nsa says fire it up

feels like the right hand is trolling the left

@kimmonismus US defense procurement cycles have historically tolerated operational use alongside supply-chain red flags—this dual-track treatment for Anthropic suggests Claude Mythos already cleared technical evaluations, but contractual or compliance friction remains.

@kimmonismus ah yes, the classic dont talk to me or my son again energy

@kimmonismus Classic government split brain: one agency blacklists for supply chain risk, another uses it anyway. If this is true, what’s the real control point, model access or procurement paperwork?