Last Wednesday, Chipotle Mexican Grill reported strong earnings against the backdrop of an ugly food-safety news cycle: $3.35 billion in revenue, up 9.3 percent, and a raised full-year outlook. The cyclospora outbreak tied to Taylor Farms lettuce at Taco Bell still cost Chipotle about two points of sales in the back half of July because people were skipping the burritos everywhere. CEO Scott Boatwright told analysts he wanted to "put that to rest": "We're not impacted, or we are impacted from a sales perspective, but we're not involved in the Cyclospora conversation today. The products that are caught up in that conversation, we don't use on our menu, and our lettuces are sourced here in California."

Boatwright did everything right. He named the specific outbreak, he scoped the denial. Every word was accurate.

But then yesterday, Minnesota health officials said they're investigating a separate salmonella outbreak, suspected, not yet confirmed, to be linked to jalapeños served at Mexican-style, quick-service restaurants. Of 110 Minnesota cases, officials interviewed 84 people, and 75 of them had eaten at a Chipotle between mid-June and mid-July.

Chipotle pulled the jalapeños, and the stock dropped about 9 percent in afternoon trading.

Now look at the statement Chipotle put out yesterday, because it's the opposite of the play you watched Taylor Farms run into a wall last month.

By yesterday afternoon, Minnesota's senior foodborne-disease epidemiologist, Carlota Medus, told reporters that "based on all the evidence so far, the food that made people sick was served at other restaurants as well" and that Chipotle "has fully collaborated on every aspect of this investigation." Per Medus, the department isn't worried about Chipotle right now.

That deserves to be lit up in neon lights: a state health official, mid-outbreak, vouching for a company under investigation. The cooperation and partnership earned that -- a traceability system that gave investigators the suspected lot, the growers, and the exact restaurants instead of making them dig. Chipotle is playing this correctly, and the people who matter are saying so out loud.

Expect more of these tests. Federal outbreak investigations are running at their fastest pace in at least five years, meat and poultry recalls have already blown past the recent annual average, and the FDA is juggling four separate cyclospora outbreaks this year all while budget cuts have thinned the surveillance systems that catch these things early. Experts disagree on whether the system is measurably worse yet, but nobody disputes that the caseload is up and the watchdogs are stretched. That math means more companies getting the call Chipotle got, with less official air cover to go around.

Over the weekend, OpenAI ran what TechCrunch reports was its first-ever influencer brand trip, a luxury retreat in upstate New York billed as "Summer Club," with farm-to-table dinners, beekeeping, and classes on its newest tools. The posts were exactly what you'd expect: pretty, warm, sponsorey-vibes. The comments were decidedly not.

The trip took place the week after reports that OpenAI is nearing a deal to lease a $500 billion data center complex in Ohio, and commenters connected those dots fast. One widely shared post: "oh how nice of chatgpt to pay for influencer brand trips to the woods before they destroy what's left of the natural world." Others brought up the $200 million Pentagon contract the company won last year. TechCrunch found commenters insinuating one attendee had sold her soul "for a nice hotel room" and asking another for a get-ready-with-me touring a data center.

Then came the part I've been watching on Instagram since Monday: attendees answering their critics themselves, and mostly badly -- snapping back, getting defensive, treating substantive questions about data centers and energy like personal attacks. One creator appears to have deleted her video from the trip while keeping the posts that don't reference it. Another wrote on LinkedIn that she thought people "would be interested in the behind-the-scenes tips and tricks" and "didn't for one second consider it would be controversial." These are people whose entire professional skill is communicating enthusiasm. Nobody brought them there for adversarial Q&A, and if anyone prepped them for pushback, it doesn't show but what did show was defensiveness, petty personal attacks, and anger.

OpenAI's statement, given identically to BuzzFeed and TechCrunch: "We welcome healthy debate as AI becomes more prevalent, and we value creators who choose to engage with us, attend our events, ask tough questions, and learn alongside everyone else." Catch the irony because OpenAI is praising creators for asking tough questions during a weekend when the tough questions were being asked OF the creators, who had no answers. To be fair to the comms team, the statement itself isn't the problem here and a company wading into its influencers' comment sections to fight their battles would have been worse. The statement is just where you can plainly see that the program had a content plan but no backlash plan.

The standard creator brief covers FTC disclosure, content approvals, and posting windows but most don't cover what to say when a stranger asks a hard but fair question about the company whose retreat you just posted from. And this category of trip isn't going away: Microsoft reportedly took Copilot creators to the Super Bowl in February, Anthropic has reportedly hosted an influencer dinner for Claude, and OpenAI hired Instagram's Charles Porch in February as its first VP of global creative partnerships. So the programs are built but the crisis half isn't.

The fix here wasn't complicated and it isn't media training. I think for an issue / organization this complex and in a space this heated, OpenAI should have gotten the group in a room for an hour at the beginning of this trip and walked through the top questions they're probably going to get in their comment section. You can offer up supporting facts or the company's answer to each in words a human would say out loud. You could give them a name at the company to send anything harder, and explicit permission to use it: "good question, not mine to answer" is a fair response for someone who is not an approved spokesperson for the company. And decide together, before the first post goes live, what happens in the comments. What deserves a reply, what gets left alone, and when should something get deleted if at all. None of this turns a creator into a spokesperson, it just stops you from putting someone in front of your hardest questions with less preparation than you'd give a summer intern answering the office phone.