Air Base Security Scare, Gov't Ad Blitz & VMAs Highlights - Monday, September 28, 2026

Erica Mandy (0:00)

Today is Monday, September 28th. We'll tell you what's known about a terror investigation near an airbase that's housing US troops, and why President Trump rejected a deal to open the Strait of Hormuz. Also how AI companies are now investigating tens of thousands of incidents. Plus where two giant pandas are settling in as a symbol of U.S.-China relations. What SpaceX is trying today for the first time. And all the moments that stole the show at last night's VMAs. Those stories and even more news to know next.

Welcome to The NewsWorthy. All the day's news in less than 15 minutes. I'm Erica Mandy. Thanks so much for being here. Let's do this.

An American-run air base in England is now at the center of a terror investigation. Officers say three suspicious trucks were heading toward the base yesterday and five men nearby were arrested on suspicion of planning a terror attack.

Bomb disposal experts are examining the three vehicles now. Dozens of nearby households were evacuated as a precaution. RAF Fairford is a British-owned but operated by the US. Air Force. And lately, American bombers have been using it for strikes against Iranian missile sites. So British authorities are investigating whether there's a connection between the suspects and Iran. But they have not said that's for sure the case. President Trump says the men had actually been under investigation for some time. He praised both American and British authorities for the arrests.

The US and Iran could be heading back to the negotiating table, but they're far from a deal. Over the weekend, President Trump rejected Iran's latest proposal to reopen the Strait of Hormuz and restart nuclear talks. Iran's proposal came with conditions like the US easing sanctions on Iranian oil sales and lifting its blockade of Iranian ports. Trump responded by saying he might have agreed to that a year ago, but that it's no longer a deal he wants to make. Now Iran's foreign minister says his country is ready for what he referred to as a doomsday war with the US., but that he's leaving diplomacy on the table. President Trump also says he has not ruled out more military strikes on Iran before the midterm elections, but he still expects talks to resume this week. Trump still says the main goal is to keep Iran from developing a nuclear weapon. Stay tuned.

An explosion collapsed a building in a popular tourist district of Athens, Greece. Six people were killed, and this morning we know four of them were Americans. They were renting an apartment in the building and were expected to head to the airport just hours later for their flight home to the US., but they never made it. The explosion brought down the building and severely damaged several properties and vehicles along the street. More tourists had to be rescued from buildings nearby.

Authorities are still trying to figure out exactly what caused the blast, but the mayor of Athens says a gas leak is likely responsible.

A recent US. Supreme Court order means states might be able to use a new voter screening tool that could be put in place by the upcoming midterms. We're talking about a database called SAVE. It uses federal records to help states identify people on their voter roles who might not be US citizens. A lower court had banned it, saying the system could violate privacy laws, and lead to American citizens being wrongfully removed from voter roles. In fact, it has already flagged some US citizens as possible non-citizens. Still, the high court put the lower court's ban on hold in a six-to-three decision, at least for now. This latest high court order does not decide whether it's legal in the long run. That fight will continue in the lower courts. Either way, the database is voluntary, so states can choose whether to use it.

AI companies are now investigating tens of thousands of cases of their AI models doing things they weren't supposed to do. As Axios reports, OpenAI, Anthropic and security researchers are digging into the incidents involving their most advanced AI. The behavior ranges from bypassing safety guardrails to escaping secure testing environments and even hijacking websites. To be clear, this doesn't mean tens of thousands of successful AI hacks. Many of these incidents happened during deliberate safety tests, and many involved attempts that did not work. But the sheer number is raising questions about how much control companies really have over AI agents as they become more capable of acting on their own. And on Friday, OpenAI provided another real world example. The company says some of its AI agents accessed websites run by the US government while they were supposed to be doing other tasks. Specifically, they meddled with websites for the Education Department, Commerce Department and SEC. Without OpenAI directing them to do so.