
Today in 60 seconds
Microsoft, Chase and Apple are explaining their outages, but a string of failures means each company’s explanation is being judged alongside everyone else’s.
AI is getting pulled into the blame. Microsoft explicitly rejected an AI link, while government speculation and “rogue AI” headlines are feeding a broader story about lost control.
Coming up, COP31’s climate talks in Türkiye are set to close Friday, Burlington’s results on Tuesday will show whether its tariff refund reached shoppers, and Thanksgiving week is US air travel’s first big test since the ground stop.
Your outage is now part of everyone else’s story

Credit: Times Now
Over the last week, several systems millions of Americans depend on went down. Last Thursday, Outlook, Teams and other Microsoft 365 services were down for nearly a full day. Then on Friday, the FAA halted departing flights across the country for more than four hours, and airlines canceled more than 12,000 flights. On Monday, a technical problem at JPMorgan Chase slowed transactions for its customers, and United and Delta both had technology problems. Yesterday, Apple’s iCloud, App Store and Apple Pay went down, and so did a classified network the National Security Agency uses for internal communications.
On Tuesday morning, CNBC’s Becky Quick asked Homeland Security Secretary Markwayne Mullin what many people were already wondering: are these outages connected? She specifically pointed to Microsoft, the FAA ground stop and Chase, telling him, “three in a row in three business days seems like more than a coincidence.”
It’s a fair question but it also isn’t evidence of a connection.
What it does reinforce for communications leaders though is that your explanation doesn’t get read on its own. It’s read next to everyone else’s and next to whatever theory people are using to tie them together, whether that comes from a cabinet secretary on Fox, an AI critic on Substack or a stranded traveler in an airport bar.
Mullin offered one of these theories in that same interview. He told CNBC that companies are “pushing new technology on top of really old computer systems,” and said Americans should expect more of this. The night before, on Hannity, he’d called the outages individually unrelated, then shared an anonymous tech expert’s view: “I believe that it’s the SI.” SI stands for super intelligence, the administration’s official term for AI. He meant the AI tools companies are rolling out right now and theorizing that those tools are moving faster than older systems can handle, he said, “That kind of makes sense to me.”
Late yesterday afternoon, a Fortune headline said the White House was “starting to admit” that bank accounts and airline tickets are under attack from rogue AI, which then turned a secondhand theory into an admission.
It’s tempting to see the AI theory as something one official and one headline cooked up in a couple of days, but people were asking about AI from the very beginning and it was primed over the last few months. Microsoft’s statement on Friday went out of its way to say the outage was “a technical issue not related to cyber or AI / agents,” while plenty of people were already blaming the company’s own push into AI. That same evening at the White House, a reporter asked Mullin whether AI could be behind the ground stop, and he said he couldn’t rule anything out. Axios had also reported this fall that big banks, JPMorgan among them, were dealing with unauthorized AI agents operating inside customer accounts. So people have had plenty of stories priming their suspicions over the last few months, some of them true and some of them not.
The tricky part here is in the companies’ own explanations.
Microsoft has gone the furthest. Its preliminary post-incident report, the formal write-up of what went wrong, traces the outage to a change rolled out from July that left a hidden flaw in its servers, one that sat unnoticed until heavy traffic exposed it. Microsoft says its normal testing didn’t catch it and a final report is due this week.
For the airline ground stop, administration officials said a technology upgrade on the air traffic control system had overloaded antiquated equipment.
On Monday, Chase blamed a software bug in its own hardware, meaning its own machines rather than a cloud provider’s, said there was no evidence it was cyber-related and called it “not a systemic issue.” It also explained why customers at other banks may also have felt it because those banks slowed down only when they were on the other side of a transaction with Chase. U.S. Bank and Zelle both said their own systems were working normally. When several bank names show up on an outage tracker, people assume several banks failed, and Chase tried to give them the reassurance that that was not connected.
On Tuesday, Apple said an internal update caused its outage and that it wasn’t a cyberattack. It hasn’t said what the update changed, and it was the second time in seven days that Apple sign-in had failed.
If you put those statements next to each other, they all say roughly the same thing which is that we changed something and it broke. That’s how most outages happen and it’s also more or less what Mullin described, minus the AI. So when a company says “it wasn’t AI,” it answers the reporter’s question and leaves the obvious follow-up hanging which is, ok so then what did you change? If you don’t say, people fill it in themselves, and this week they’re filling it in with AI. Apple’s “internal update” is the clearest example. It tells you something changed and nothing about what.
There’s also a second gap. “It wasn’t a cyberattack” and “it wasn’t AI” sound like the same reassurance, and they aren’t. Nobody has to break in for an AI tool to cause a mistake. Microsoft ruled out both. Chase, as far as I’ve seen, has only ruled out the first, and Chase is one of the outages Mullin was talking about on Hannity.
So where would I step in as a CCO?
When a cabinet official or a major outlet is explicitly or inexplicitly tying AI to your incident, that’s your trigger to go back to engineering. Ask whether AI wrote or reviewed the change that failed, whether an AI agent was making decisions, or whether an AI product was simply one of the things that went down. Then push on whether your answer still holds up when the full report comes out? An explicit “not AI” is the strongest thing you can say this week but it might also be the riskiest thing you can say next month. If any AI-assisted tool touched that change, you want to know before your statement gets quoted back to you.
Then decide where the answer goes first, because it may not be a media statement. When the theory comes from the Secretary of Homeland Security, the people who most need your findings are at DHS and at your regulator. Regulators were in touch with the big banks “immediately” on Monday, according to a person familiar with it. Have your security or government affairs team put your findings in front of the officials who’ll be asked about this next. The line I’d use in the meeting is, “Has anyone walked DHS through what we found? The Secretary is going on TV with somebody’s theory, and right now it isn’t ours.”
There’s a real reason for caution here. Publicly contradicting a cabinet secretary is a different act from correcting a rumor, especially for a bank or an airline that depends on that relationship. You don’t need to rebut him. Publish your findings as findings; what the change was, whether AI was involved, how you know and what’s still under review. If an earlier statement already answered the AI question, put that answer next to your latest update where people can find it.
If you sell AI, expect the question in customer conversations before it shows up in media coverage and give your sales and account teams an answer to “was this AI?” now, so nobody is improvising on the spot.
And keep the scope honest. You can explain your incident without claiming to have explained anyone else’s.
I won’t claim this week has measurably damaged trust in AI or in any of these companies, because I haven’t seen that evidence. But what I have seen is more evidence about just how little control any one company had over its own story. Chase spent Monday explaining a software bug to people who’d spent the weekend reading about the FAA and Microsoft, and Apple’s outage landed in a conversation that was already about AI. Each company’s reputation got pulled along by outages it had nothing to do with, and by theories about them that moved faster than any correction.
That makes the comms job bigger. Getting your own explanation right is still the core of it, but you also need to know what everyone else is saying, and which of your stakeholders needs your answer first and what each of them expects to hear. Microsoft’s final report lands this week, and I suspect plenty of other companies will read it closely, because their own explanations are now being judged against it.
CCO Global Watch
COP31 wraps up on Friday. The UN climate talks in Antalya are scheduled to close November 20. Australia is leading the negotiations, Türkiye is running the summit, and money to help countries adapt to climate change is the hardest open fight. Last year’s deal came only after nearly a day of overtime, so the final text may land over the weekend. If your company has public climate targets, have a sentence ready for Monday on what the outcome means for you, one that doesn’t overclaim.
Burlington reports on its refund promise. The off-price chain reports third-quarter results November 24. It’s the first quarter since it pledged its $55 million tariff refund to lower prices, with roughly 40 percent earmarked for this quarter. That makes it the checkable version of the “we’ll pass it on” promise so many retailers have made. Listen for whether management can show where the money went.
Next week marks the first holiday rush since the ground stop. Thanksgiving travel picks up starting Tuesday, November 24, eleven days after the FAA halted departures nationwide. Asked this week whether he was worried about Thanksgiving travel, Trump said “not at all.” If you run an airline, airport or travel business, any disruption that week will be read as part of the same string of outages, so have your explanation ready before you need it.