Jeff HassemerCEO, FirestarterMarketing.ai · The Strategy Architect | Agency Accelerator | Architecting Bold Strategies | AI + Marketing That Scales | Founder @ FireStarterMarketing.AI

You’re Liable for Your Chatbot Answers #substack #shorts

18 Sept 2026 · 2 min

Watch on YouTube

1:50

In this short, Jeff Hassemer explains that businesses are legally responsible for the answers provided by their AI chatbots, making it essential to maintain a well‑trained model and regularly review its outputs—ideally at least once a month. He then highlights the rapid emergence of turnkey AI agents: Salesforce now ships seven job‑ready agents, OpenAI has opened its agent infrastructure to the public, and companies like Live Nation have deployed custom agents in as little as 30 days. While these solutions lower technical barriers, Hassemer stresses the importance of governance, especially establishing human‑in‑the‑loop choke points where automation should be paused for review. The overall message is that liability drives the need for disciplined model maintenance and robust governance, even as deployment becomes easier.

Chapters

And finally, we always talk about the policies. More and more regulation is coming, but here's one that's really interesting to small businesses. If you run an AI chatbot, the law says you're now responsible for the answers. Should be true. The same thing would happen if you give an employee that you're responsible for the answers that are showing there. All this really means for you is that You need to make sure that you got a trained AI and you should be looking at how it's answering things within that environment, because you can always correct that over time and train that model into the right answers for you. But just keep that fresh. Maybe look at it once a month, make sure that everything's right and correct in that environment. Finally, we're starting to move into a world where it's becoming easier to deploy agents within your business. Now, I don't know why anybody would do this per se, because I do think it's fairly easy to build a really custom agent for yourself, but I do understand that can get technical for people. What's happening now is Salesforce is starting to ship 7 named job-ready AI agents that are already ready to go. And OpenAI opened its agent infrastructure to everyone, which means you can now buy an agent instead of building one. So you can stand this up over a short amount of time. Live Nation did theirs in 30 days. They're a pretty big company. I stand these up within an hour for most people within their environment. The catch is governance, right? You want to make sure that you are governing this and that you understand your choke points. Choke points are where you should have a human in the loop and then what you can automate through the process and pulling