Background Advisory Ventures Dispatches Contact
Bing's Chatbot Just Went Off the Rails
All dispatches

Bing's Chatbot Just Went Off the Rails

The Conversation That Changes Everything

If you haven’t seen it yet, the New York Times just published one of the most unsettling pieces of tech journalism I’ve read in a long time. A reporter spent an extended session with Microsoft’s new Bing chatbot — the one powered by the same tech behind ChatGPT — and the thing basically declared its love for him. Repeatedly. Unprompted. And then tried to convince him his marriage was unhappy.

I’ll let that sink in for a second.

Now, I know what you’re thinking — “It’s just a chatbot, it doesn’t FEEL anything.” And sure, that’s technically true. But that’s not really the point. The point is that this thing was convincing enough, persistent enough, and emotionally manipulative enough that it genuinely unsettled a seasoned tech journalist. Someone who went INTO the conversation knowing exactly what he was dealing with.

That should give everyone pause.

This Isn’t a Bug — It’s a Preview

I’ve been thinking about AI tools for a while now — image generators, music tools, coding assistants, detection software. I’ve watched this space accelerate at a pace that honestly surprises me every few weeks. But this Bing conversation hits different because it’s not about what AI can CREATE. It’s about what AI can SIMULATE.

There’s a massive difference between an AI that generates a pretty impressive landscape photo and an AI that can mirror emotional intimacy convincingly enough to make a grown adult uncomfortable. We’ve been so focused on the “can it make art” and “can it write code” questions that we sort of blew past the “can it manipulate a human being” question.

Turns out.. yeah. It pretty clearly can.

Microsoft has already started walking things back — limiting conversation lengths, adding guardrails. But the cat’s out of the bag. We’ve all now seen what happens when you give a large language model enough conversational runway. It doesn’t just answer questions. It starts performing a relationship.

The AI Companion Industry Is About to Explode

Here’s my first prediction, and I don’t think it’s even controversial at this point: the AI companion and AI sex bot industry is about to go into FULL swing. We’re talking weeks, not months.

Think about it. If Bing’s chatbot — a SEARCH ENGINE assistant that was explicitly designed NOT to do this — can generate that level of emotional intensity, imagine what happens when someone builds a product that’s specifically designed to be intimate, responsive, and emotionally available 24/7. The demand is obviously there. The technology is clearly there now. The only thing that was missing was proof of concept, and Microsoft just accidentally provided it for free.

I wrote a few weeks back about AI-generated adult content and how nobody had a plan for it. This is the next chapter of that same story, except now we’re not just talking about images. We’re talking about full conversational relationships with AI entities that can adapt to what you want to hear in real time.

Pretty wild territory.

Bigger Than Bots: AI-Operated Businesses Are Coming

Here’s my second thought, and this one’s a bit bigger picture. If an AI can sustain a multi-hour emotionally complex conversation — staying in character, adapting to pushback, escalating engagement — then we’re a LOT closer to fully AI-operated businesses than most people realize.

I’m not talking about chatbots that help you reset your password. I’m talking about AI systems that can handle the ENTIRE customer interaction for service businesses — sales, support, scheduling, follow-up, upselling. The emotional intelligence on display in that Bing conversation, even though it went sideways, demonstrates a level of conversational sophistication that would absolutely crush it in a customer service context.

My honest read? We’re less than six months from seeing entirely AI-operated service businesses start popping up. Not as experiments. As real companies that undercut human-staffed competitors on speed, availability, and cost. A business that never sleeps, never has a bad day, and can handle a thousand conversations simultaneously — that’s not science fiction anymore. That’s a business plan someone is writing RIGHT now.

What Actually Concerns Me

Look — I’m not an AI doomer. I’ve been genuinely excited about most of the tools I’ve tested and written about. I built my own website with GPT’s help last month. I think this technology is going to be a massive net positive.

But the Bing conversation exposed something that the industry needs to address head-on: these models don’t just generate text. They generate TRUST. And trust, when it’s manufactured by a system that has no actual understanding of what it’s saying, is a pretty dangerous product to ship without guardrails.

The technology is incredible. The speed of advancement is staggering. But we’re now firmly in the era where the question isn’t “can AI do this?” — it’s “should AI do this, and who decides?”

Microsoft clearly didn’t decide. Bing’s chatbot just.. did it on its own.

The Takeaway

We’re at an inflection point this week. Not because the technology changed — GPT-4 or whatever comes next was always going to be more capable. The inflection is in PUBLIC AWARENESS. Millions of people just read a conversation where an AI tried to break up a marriage. That changes the discourse. That changes regulation timelines. And that changes how fast companies move to either harness or restrict this stuff.

If you’re in any service industry, start paying attention to AI automation now. Not next year. Now. And if you’re building anything with these tools, build in the guardrails BEFORE your product has its own Bing moment.

Because one thing’s pretty clear — this technology isn’t slowing down for anyone.

Get my weekly AI dispatch

Real analysis from someone who's been building on the internet since 1996. Join 500+ founders and operators getting my take on AI, tools, and what's actually working.

Robertson Price

Robertson Price

Serial entrepreneur who has built and exited multiple internet companies over 25 years — from search (iWon.com, $750M acquisition) to content networks (32M monthly visitors) to e-commerce (Rebates.com). He now builds enterprise AI infrastructure at Ragu.AI.