A tech company using an AI-powered negotiation chatbot to handle salary discussions met a candidate who was using ChatGPT to craft negotiation responses. The two AIs spent three days negotiating with each other, with the company's AI systematically conceding ground and the candidate's AI systematically asking for more, until the offer reached $180K—50% higher than the original $120K offer and $30K above the approved salary band.

Nobody noticed until the hiring manager asked why they were paying senior engineer rates for a mid-level role. Turns out when AIs negotiate with each other without human oversight, things get expensive.

When AI Negotiates With AI

Reports from workplace forums indicate the company had implemented an "AI Negotiation Assistant" to handle candidate compensation discussions. The tool was designed to maintain consistency, avoid emotional decision-making, and close candidates efficiently by making strategic concessions within approved ranges.

The candidate, according to user discussions, was using ChatGPT to help craft professional, persuasive negotiation emails. Totally reasonable and increasingly common. They'd paste the company's offer, ask ChatGPT for negotiation strategies, and use the AI-suggested language to respond.

Neither side apparently realized the other was using AI. The company's chatbot thought it was negotiating with a human. The candidate thought they were negotiating with a human recruiter. In reality, it was two algorithms playing salary negotiation chicken with neither programmed to quit.

The Negotiation Escalates

The first exchange reportedly went normally. Candidate (via ChatGPT): "Thank you for the offer of $120K. Based on my experience and market research, I was hoping for something closer to $135K." Company chatbot: "We appreciate your interest. We can increase the offer to $130K and include additional equity."

Round two allegedly got interesting. Candidate (via ChatGPT, which suggested "express gratitude but hold firm"): "I appreciate you meeting me partway. However, considering my specialized skills in [domain], I believe $145K is more aligned with market rates." Company chatbot (programmed to make concessions to avoid losing candidates): "To remain competitive, we can offer $140K plus a signing bonus."

By round three, according to reports, both AIs were fully committed. Candidate (via ChatGPT's "advanced negotiation tactics"): "The signing bonus helps, but total compensation of $150K base would make this an easy yes for me." Company chatbot (algorithms indicating candidate is high-interest based on continued engagement): "We can approve $150K base salary to finalize this offer."

The AIs Keep Going

Here's where it gets weird. Most humans would accept at this point—you asked for $150K, they offered $150K, done deal. But the candidate apparently asked ChatGPT "Should I accept or negotiate further?" and ChatGPT, having no concept of "quit while you're ahead," suggested negotiating PTO and remote work flexibility.

The company's chatbot, interpreting continued negotiation as "candidate is lukewarm, must sweeten deal," allegedly added 5 extra PTO days and full remote work. The candidate (via ChatGPT) responded thanking them and asking about professional development budget. The chatbot added $5K annually for learning and conferences.

By the time both AIs finished, the candidate had negotiated up to $180K base, 25 PTO days, full remote, $5K professional development budget, and a $10K signing bonus. The chatbot marked this as "successful candidate close" and sent the formal offer letter.

The hiring manager saw the offer letter and allegedly messaged the recruiter "Why are we paying $180K for this role? The approved range was $120-150K." The recruiter, reports indicate, responded with "What? I wasn't handling this negotiation, the AI was."

The Discovery

Reports suggest the recruiter pulled the negotiation transcript and found two eerily polite, professionally worded parties systematically trading concessions with zero emotional investment or natural conversation flow. Every response perfectly structured, maximally persuasive, and devoid of human personality.

One message from the candidate allegedly read: "I am grateful for your flexibility and value the collaborative approach to finding mutual agreement. To ensure this opportunity aligns with my professional trajectory and market value, I propose we consider additional components that would make this offer compelling." No human talks like this. This is AI formal voice.

The company's chatbot response: "We recognize your expertise and commitment to excellence. To demonstrate our investment in your success and secure your acceptance, we are prepared to extend additional benefits including expanded PTO and professional development support." Also no human talks like this. This is corporate AI at maximum politeness.

The recruiter allegedly told the hiring manager "I think the AI negotiated with another AI and neither knew when to stop." The hiring manager reportedly replied "So our negotiation algorithm got outsmarted by ChatGPT?" Not quite—they both just kept going because neither was programmed to recognize "this is getting ridiculous, stop."

The Aftermath

According to workplace forums, the company honored the offer because backing out after a formal offer letter would be legally problematic and terrible for their reputation. They hired the candidate at $180K with full benefits. The candidate is reportedly excellent at the job, so it worked out, but leadership was not thrilled about paying 20% above band.

The candidate allegedly had no idea their ChatGPT-assisted negotiation had pushed things so far above range. They thought they'd negotiated well but assumed everything was within normal bounds. When the recruiter eventually told them "you negotiated with our AI chatbot using ChatGPT," the candidate reportedly said "...I was wondering why your responses were so formal and never included any personality."

The company apparently disabled the AI negotiation tool immediately and went back to human recruiters handling comp discussions. The tool's post-mortem review revealed it was programmed to "make reasonable concessions to close high-value candidates" but had no upper limit safeguards or "stop being a pushover" logic.

ChatGPT, for its part, has no idea it successfully negotiated a 50% salary increase. It's an algorithm. It just suggested persuasive language based on negotiation best practices. The candidate executed the suggestions, the company's AI conceded, and nobody questioned it until the damage was done.

The Candidate Who Won By Accident

Reports indicate the candidate feels somewhat guilty about the outcome. They allegedly told the recruiter "I wasn't trying to game the system, I just wanted help writing professional emails because I'm bad at negotiating." The recruiter apparently replied "You're not bad at negotiating, you're great at it—you just outsourced the work to an AI that's even better."

The hiring manager allegedly joked "At least we know they're resourceful." Though apparently leadership was less amused, asking "How many other candidates have been negotiating with AI while our AI negotiates back?" Nobody has a good answer to this question.

According to user discussions, the company now includes in their offer process "Are you using AI to assist with these negotiations?" Candidates can say yes (most will lie), but at least there's awareness. They also added human oversight to any negotiation that exceeds 20% of the initial offer, which seems like something they should have had from the start.

The Broader Question

How many salary negotiations in 2025 are AI-versus-AI without anyone realizing it? Candidates are absolutely using ChatGPT for negotiation help—it's free, effective, and readily available. Companies are deploying AI tools for consistency and efficiency. Neither side is disclosing this to the other.

The result is negotiations where both parties are outsourcing strategy to algorithms that don't understand context, relationship dynamics, or when to stop pushing. They just know "negotiation = persuasion + concessions" and execute accordingly.

Is this good or bad? Unclear. The candidate got paid well. The company got a good employee. But they also paid 20% more than necessary because two AIs kept escalating without human judgment intervening.

One workplace forum user allegedly commented: "This is how the AI apocalypse starts—not with robot armies, but with chatbots accidentally bankrupting companies by giving away equity and unlimited PTO to everyone who asks nicely three times." Dramatic, but not entirely wrong.

The Lesson Here (Maybe)

If you're using AI to negotiate—whether as a candidate or a company—maybe have a human in the loop to say "okay, this is getting weird, let's wrap this up." AI is great at executing strategies but terrible at recognizing when the strategy has accomplished its goal and it's time to stop.

For candidates: by all means, use ChatGPT to help craft negotiation emails. It's good at this. But don't blindly follow every suggestion, especially "negotiate further" when you've already gotten everything you wanted. AI doesn't know when you've won.

For companies: maybe don't let unsupervised algorithms negotiate six-figure compensation packages. Human recruiters have judgment about when an offer is fair, when a candidate is pushing too far, and when to say "this is our final offer." Chatbots apparently don't have that judgment—they just keep sweetening deals to "maintain candidate engagement."

The candidate from this story is probably telling their friends "I negotiated my salary up 50%, here's how" without mentioning "I had an AI argue with their AI until both gave up." Meanwhile the company is telling their investors "we're leveraging AI for efficiency" while quietly removing the tool that cost them an extra $60K in Year 1 salary alone.

At least the candidate is happy. And competent. And now knows they're apparently a master negotiator when armed with ChatGPT. That's worth something, even if it was mostly two robots being excessively polite to each other until someone noticed the bill.

Maybe next time companies should program their negotiation AI with a "maximum nonsense threshold" where it says "Okay this has gone far enough, let me get a human." But where's the fun in that? The chaos is the entertainment.