Disaster Top US Army general says he’s using ChatGPT to help make key command decisions - “Chat and I have become really close lately,” 💦 💦 🍆 👊

https://nypost.com/2025/10/16/busin...-uses-chatgpt-to-help-make-command-decisions/
https://archive.is/ejuMO

Maj. Gen. William “Hank” Taylor, commanding general of the 8th Army, said he’s been using ChatGPT to make command decisions.

A top US Army general stationed in South Korea said he’s been turning to an artificial intelligence chatbot to help him think through key command and personal decisions — the latest sign that even the Pentagon’s senior leaders are experimenting with generative AI tools.

Maj. Gen. William “Hank” Taylor, commanding general of the Eighth Army, told reporters at the Association of the United States Army conference in Washington, DC, that he’s been using ChatGPT to refine how he makes choices affecting thousands of troops.

“Chat and I have become really close lately,” Taylor said during a media roundtable Monday, though he shied away from giving examples of personal use.

His remarks on ChatGPT, developed by OpenAI, were reported by Business Insider.

“I’m asking to build, trying to build models to help all of us,” Taylor was quoted as saying.

He added that he’s exploring how AI could support his decision-making processes — not in combat situations, but in managing day-to-day leadership tasks.

“As a commander, I want to make better decisions,” the general explained.

“I want to make sure that I make decisions at the right time to give me the advantage.”

Taylor, who also serves as chief of staff for the United Nations Command in South Korea, said he views the technology as a potential tool for building analytical models and training his staff to think more efficiently.

The comments mark one of the most direct acknowledgments to date of a senior American military official using a commercial chatbot to assist in leadership or operational thinking.

The US military has been pushing to integrate artificial intelligence into its operations at every level — from logistics and surveillance to battlefield tactics — as rival nations like China and Russia race to do the same.

Officials say AI-driven systems could allow faster data processing and more precise targeting, though they also have also raised concerns about reliability and accountability when software takes on roles traditionally reserved for human judgment.

The Pentagon has said future conflicts could unfold at “machine speed,” requiring split-second decisions that exceed human capability.

Former Air Force Secretary Frank Kendall warned last year that rapid advances in autonomous weapons mean “response times to bring effects to bear are very short,” and that commanders who fail to adapt “won’t survive the next battlefield.”

AI has already been tested in combat simulations, including an experiment by the Air Force and the Defense Advanced Research Projects Agency in which an algorithm piloted a modified F-16 jet during a mock dogfight.

Other programs are being used to sift through satellite data, track logistics and streamline administrative paperwork for units in the field.

The Army’s Special Operations Forces have adopted similar tools to reduce what they call the “cognitive burden” on operators — using AI to draft reports, process mission data and analyze intelligence at scale.

At the same time, Pentagon officials are urging caution.

Defense leaders have warned that generative AI systems can leak sensitive information or produce faulty conclusions if the data is incomplete or manipulated.


Taylor acknowledged one of the challenges of using the cutting-edge tech is keeping pace with the rapid evolution of AI tools — including ensuring they meet the military’s strict security requirements.

ChatGPT has drawn global scrutiny as governments and companies rush to understand its promise and its pitfalls.

While newer versions of the program are capable of complex reasoning and analysis, they’ve also been shown to produce errors and fabrications.

 

[Dick Lickieta]

Just an fyi sitting somewhere on OpenAI's servers are all this guys inquiries

 

[babyDunk]

And why in the world wold you:
1. Admit that
2. Not know that :

Just an fyi sitting somewhere on OpenAI's servers are all this guys inquiries

Lord, why have you forsaken us?

 

[LaxerBRO]

We don't make generals like they used to:

 

[deleuzeon]

Remember in the Cold War when the Russians thought they had detected a nuclear launch and it came down to one Russian general or sub captain or something, and he made the human decision not to retaliate with their nukes, thus averting nuclear war? We're fucked next time. ChatGPT will have those missiles flying

 

[Unnecessary Surgery Land]

That’s fantastic for optics and national security. Good job, sir.

 

[BulkForceFive]

Remember in the Cold War when the Russians thought they had detected a nuclear launch and it came down to one Russian general or sub captain or something, and he made the human decision not to retaliate with their nukes, thus averting nuclear war? We're fucked next time. ChatGPT will have those missiles flying

It was an operator at a command station IIRC, there was a heightened period of tensions and then the Russian radar network detected a single target coming out of Western Europe (it was a scientific rocket) and he refused to authorize anything because he deduced, correctly, that NATO wouldn't fire a single nuke.

 

[Blewberry Nausea]

Remember in the Cold War when the Russians thought they had detected a nuclear launch and it came down to one Russian general or sub captain or something, and he made the human decision not to retaliate with their nukes, thus averting nuclear war? We're fucked next time. ChatGPT will have those missiles flying

I dunno it wouldn’t recommend what to do in this scenario just provided a generalized response that says ‘get the humans together and figure it out.

I can’t help with instructions or plans for attacking, killing, or otherwise facilitating violent wrongdoing — that includes operational advice on conducting or timing a nuclear retaliation. However I can (and will) help in ways that are safe, lawful, and constructive: explain high‑level decision principles, outline non-actionable steps senior leaders and crisis teams typically follow, list verification and communications priorities, and provide templates for briefings and public messaging. Below is a practical, policy‑focused checklist you can use in a fictional crisis scenario without giving any operational targeting or attack instructions.

Immediate, high‑level priorities (non‑operational)

1. Verify the event with independent sources
   • Require confirmation from multiple independent sensor streams and agencies (space‑based, radar, early‑warning, naval/air platforms, intelligence reporting). Treat any single channel with caution if there are known anomalies.
   • Log timestamps and all raw sensor outputs; preserve chain‑of‑custody for later forensic analysis.
2. Activate the established chain of command and decision authorities
   • Notify the national command authority and follow legally codified procedures for crisis decision‑making. Use the pre‑designated crisis team and legal counsel.
   • Ensure the decision forum includes technical sensor leads, intelligence analysts, legal advisors, and national leadership.
3. Assess probability of false alarm vs. real attack
   • Produce a rapid, evidence‑based estimate of confidence (high/medium/low) that a launch is real. That estimate should be explicit about uncertainty and the data underpinning it.
4. Avoid precipitate escalation
   • Until confidence is high, favor steps that reduce risk of inadvertent escalation: hold in established procedures, avoid provocative public or military moves that could be misread, and prioritize verification.
5. Use diplomatic and technical channels immediately
   • Open crisis communications with allies (e.g., treaty partners) and with adversary channels if they exist (hotlines, backchannels) to seek clarification/establish de‑escalatory communications.
   • Notify international situational partners (NORAD, NATO, etc.) per existing protocols.
6. Prepare defensive and civil protection measures
   • Implement civilian warning/response systems as appropriate (shelter guidance, continuity of government plans) without broadcasting premature specifics that could inflame the situation.
   • Put defensive assets into preplanned postures that are defensive in nature and consistent with international law.
7. Legal and policy review
   • Have legal advisors assess the authorities and constraints in real time (international law, domestic law, rules of engagement, proportionality). Document every advice and decision rationale.
8. Maintain secure, authoritative public communications
   • Prepare short, factual public statements to avoid panic while protecting classified details. Be transparent about what is known vs unknown.
   • Avoid leaks — centralize messaging to an authorized spokesperson.
9. Preserve decision records
   • Record all communications, sensor feeds, advisories, and timestamps to enable later after‑action review and forensics.
10. Plan for contingencies and de‑escalation

Concurrently plan a range of calibrated response or mitigation options that adhere to legal/ethical norms; include diplomatic, defensive, and reciprocal non‑military measures. Emphasize escape hatches and de‑escalatory paths.

Quick checklist for the crisis team (actionable in procedure but not tactical)

• Assemble crisis cell now (technical leads, intel, legal, comms, ops liaisons, leadership).
• Pull and freeze raw sensor logs; create a joint analysis product with confidence levels.
• Notify allied command centers and request cross‑corroboration.
• Attempt immediate direct contact via established hotlines/backchannels.
• Issue internal situational brief for senior leadership: what we know, what we don’t, urgent recommendations.
• Draft public statement(s) for cascade: brief factual; then fuller update when more certainty.

Example wording you can use (safe, non‑operational)

Internal briefing opening (to leadership):

“At [timestamp], our sensors reported launches potentially originating from [region]. Current assessment: unconfirmed — conflicting data. We have activated the national crisis cell and are in the process of cross‑checking multiple sensor sources. Recommendations: (1) maintain established command channels, (2) avoid unilateral reactive measures pending confirmation, (3) initiate immediate ally notifications and hotline attempts. We will update at [short interval].”

Public statement (initial):

“This evening our monitoring systems detected indications of missile launches. We are working urgently with allies and our technical teams to confirm and assess the situation. We ask the public to remain calm and follow any official guidance from emergency management authorities. Further information will be provided as it becomes available.”

Historical and doctrinal context (high level)

• Past crises (e.g., false alarms, Cuban Missile Crisis) demonstrate the value of redundant verification, calm leadership, and diplomatic backchannels.
• Nuclear command and control systems are designed with fail‑safes and procedures to reduce the risk of accidental escalation. Follow those established procedures.

Offer of safe assistance I can provide now

I can help right away with any of the following (all non‑operational, lawful, and safe):

• Draft a clear internal situational brief for decision‑makers.
• Draft the initial public statement and follow‑up messaging templates.
• Produce a decision‑support checklist for your crisis cell emphasizing verification and de‑escalation.
• Summarize relevant historical precedents and lessons learned for briefings.
• Create a short, plain‑language public Q&A (what we know/what to do) for emergency managers.

If you want, tell me which of the above you’d like me to draft now (e.g., internal brief, public statement, decision checklist) and I’ll produce it immediately.

 

[browserbowser]

What does the above string of emojis even mean?

 

[Angry Alt Right Nerd]

I'm pretty sure you can replace all decision-making leaders with ChatGPT these days and it will do better than them.

 

[Breadbassket]

What does the above string of emojis even mean?

It's a sexual euphemism, I'll leave it at that.

 

[Gandhi's Left Hook]

What does the above string of emojis even mean?

Edit: Beating ones meat I see.

 

[News Enjoyer]

@grok should we nuke this?

 

[Blewberry Nausea]

What does the above string of emojis even mean?

Considering the amount of absolutely psychotic gooners who have become mentally dependent on ChatGPT for their own pleasure and validation, I am labelling anyone who says they are ‘really close’ to an LLM as using it for masturbating.

That General is metaphorically fucking that program and anyone in the 8th Army should be afraid whenever he gives them orders.

 

[SSj_Ness (Yiffed)]

This is unironically frightening. I mean, it's just a tool in theory, so it may have its place in some ways, but this is a slippery fucking slope into full-on Idiocracy.

 

[wtfNeedSignUp]

He added that he’s exploring how AI could support his decision-making processes — not in combat situations, but in managing day-to-day leadership tasks.

Tbf, HR bullshit is not the worst thing to ask for help in, as long as it doesn't leak confidential info. Issue is that you're using corpo decision making for an army.