Raw LLM Responses
Inspect the exact model output for any coded comment.
Look up by comment ID
Random samples — click to inspect
G
Every moron holding an iPhone with facial recognition to open it is a moron. PS……
ytc_UgzIaEZcl…
G
My biggest concern still is the accuracy of AI. I have been trying to build a s…
ytc_UgzPJVN1n…
G
Bro what yall doing to these chat bots im just having a normal conversation 💀…
ytc_UgzNrxBXj…
G
What we have today, IS INDEED defined as Artificial Intelligence. Quit saying "A…
ytc_Ugxrjfs97…
G
Hi, I don't have a legal disability, but I do have a genetic condition for which…
ytc_UgwFsoIoI…
G
AI is powerful and I love AI, but I think humans are better (in terms of art)…
ytc_UgzbsEc-D…
G
Nahhhh that parents are useless and stupid careless parents,how can you blame AI…
ytc_UgxG7WzoF…
G
Why do you find it "creepy" that the female AI knew you wanted to lessen coffee …
ytc_UgyRtLgP4…
Comment
The Pentagon's three threats are mutually exclusive. Terminate the contract, designate a supply chain risk, AND invoke the DPA? Termination and supply chain risk both mean "we don't want you." DPA means "we're forcing you to work with us.
**This is a pressure play.**
A former DoJ-DoD liaison flagged this exact contradiction publicly, calling the supply chain risk designation "punitive" rather than legitimate. That's a pretty damning read from someone who used to work the DoD liaison role.
The Defense Production Act was designed to compel manufacturing of commercially available products during emergencies, i.e. ventilators, masks, vaccines. Anthropic's classified deployment is custom-built software tailored to sensitive government use.
That's a very different legal question. If Anthropic challenged this in court, the government would need to justify that a Cold War-era manufacturing law applies to compelling a company to remove safety guardrails from bespoke software. Untested legal territory and the government likely knows it.
The Pentagon admitted on the record they need Anthropic and they need them now. Claude is the only model on classified networks. There's no backup. So why issue a public ultimatum you might not be able to enforce, against a company you can't replace? A few possibilities...
One - this is theatre for a domestic audience. The administration has been framing AI safety as "woke" and this positions them as tough on companies that won't comply. Sacks has already laid the groundwork for that narrative.
Two - it's a negotiation tactic. Set an extreme deadline, make extreme threats, then settle for something in the middle - maybe Anthropic agrees to loosen some restrictions short of the two red lines, and both sides claim a win.
Three - it's genuinely about establishing precedent. If the government can compel an AI company to remove safety guardrails via the DPA, that's a template they can use on any tech company going forward. The specific dis
reddit
AI Responsibility
1772002695.0
♥ 3
Coding Result
| Dimension | Value |
|---|---|
| Responsibility | government |
| Reasoning | deontological |
| Policy | liability |
| Emotion | mixed |
| Coded at | 2026-04-25T08:33:43.502452 |
Raw LLM Response
[
{"id":"rdc_o792oiw","responsibility":"government","reasoning":"deontological","policy":"regulate","emotion":"outrage"},
{"id":"rdc_o798zf9","responsibility":"government","reasoning":"consequentialist","policy":"ban","emotion":"fear"},
{"id":"rdc_o7ab0zw","responsibility":"government","reasoning":"deontological","policy":"liability","emotion":"mixed"},
{"id":"rdc_o7bk48a","responsibility":"government","reasoning":"consequentialist","policy":"regulate","emotion":"fear"},
{"id":"rdc_o7c1d5v","responsibility":"government","reasoning":"consequentialist","policy":"none","emotion":"approval"}
]