Infotainment, EV Charger Exploits Earn $1M at Pwn2Own Automotive 2026

Trend Micro's Zero Day Initiative sponsored its third annual Pwn2Own Automotive competition in Tokyo this week, receiving 73 entries, the most ever for a Pwn2Own event.

"Under Pwn2Own rules, all disclosed vulnerabilities are reported to affected vendors through ZDI," reports Help Net Security, "with public disclosure delayed to allow time for patches." Infotainment platforms from Tesla, Sony, and Alpine were among the systems compromised during demonstrations. Researchers achieved code execution using techniques that included buffer overflows, information leaks, and logic flaws. One Tesla infotainment unit was compromised through a USB-based attack, resulting in root-level access. Electric vehicle charging infrastructure also received significant attention. Teams successfully demonstrated exploits against chargers from Autel, Phoenix Contact, ChargePoint, Grizzl-E, Alpitronic, and EMPORIA. Several attacks involved chaining multiple vulnerabilities to manipulate charging behavior or execute code on the device. These demonstrations highlighted how charging stations operate as network-connected systems with direct interaction with vehicles.
There's video recaps on the ZDI YouTube channel — apparently the Fuzzware.io researchers "were able to take over a Phoenix Contact EV charger over bluetooth."

Three researchers also exploited the Alpitronic's HYC50 fast-charging with a classic TOCTOU bug, according to the event's site, "and installed a playable version of Doom to boot." They earned $20,000 — part of $1,047,000 USD was awarded during the three-day event.

More coverage from SecurityWeek: The winner of the event, the Fuzzware.io team, earned a total of $215,500 for its exploits. The team received the highest individual reward: $60,000 for an Alpitronic HYC50 EV charger exploit delivered through the charging gun. ZDI described it as "the first public exploit of a supercharger".

Scary?

[JcMorin]

Is it me, or that's pretty scary, if there are so manu vulnerability found in a year, are many still hidden deep in the code?

Re:Scary?

[Powercntrl]

If the kind of EV chargers they're talking about are actually EVSEs, then no, it's not scary. EVSEs are actually just a glorified extension cord for connecting your EV to your home's mains. All the hardware that actually modulates the amperage draw and converts the voltage to what is required to charge the traction battery, is located inside the vehicle itself. The EVSE only tells the car what is available from the circuit it is connected to and engages a contactor that energizes the cord, after the car signals to the EVSE that everything seems good to go with the connection to the vehicle.

The worst you can really do by hacking an EVSE is to cause it to fail to function (which I guess would be a denial-of-service attack) - something that would be just as easy to accomplish by stealing/vandalizing the cord. I suppose you could also reconfigure it to present itself to the EV as having more amperage available than the circuit it is connected to can actually provide, but that will just result in a tripped circuit breaker.

Now on the other hand, if we were talking hacked L3 (DC fast charging) chargers, yeah, things can go very wrong when those malfunction.

Re:Scary?

[Zocalo]

I think a home charger is more likely for Pwn2Own. Winning an L3 DC fast-charger is cool and all, but it's not likely that the successful hacker is going to be able to do much more with it that use it as a garden ornament or oversized door stop, unless they offer a cash alternative, because things like DC fast chargers should absolutely be subjected to this kind of thing just as much as the typical consumer tools that make up most of the targets.

A cybersec seminar I was at last yeaar had a speaker from a major pen-testing company describing how they had got a L3 DC EV charger setup into a Faraday cage to see if it could be exploited (they didn't state which make, before you ask, presumably because they are in active use, but they did say that the charging network operator was their client, which is not necessarily the same as the manufacturer of the equipment, so I'm guessing probably NOT Tesla). Turns out these things were a comms nightmare, and despite the fact that they often have to have buried HV cables due to their location on the forecourt, they use wireless data links rather than hardwiring then via a secondary LV cable duct). This is broken down into:

Usage/payment processing. You'd expect this to be secure because of standards like PCI-DSS and because it's a glorified version of those remote card terminals you see everywhere, so a known tech that has been audited over and over, and that was mostly true - they couldn't get at the payment info - but they were still able to interfere with it and create a DoS and extract an awful lot of PII from users of the charging pods sent from their phone apps.

Management. These things often sit out on a forecourt, but there is usually also a management terminal located somewhere onsite showing status info, etc.. This proved to be woefully insecure, and they were able to send bogus data to the management console, and get it to show whatever status info they wanted, which is important because of the third data network.

Power supply regulation. When installed in a group, the chargers "chat" amongst themselves to optimise the distribution of the available power from the grid when the bank is close to maxing it out so that you can have a car arriving with an almost flat battery prioritised over one that is already 80% full, as well general management and heat regulation through redistribution of supply current so that nothing gets too hot. Turns out this was woefully insecure too.

By the culmination of their exercise, they were able to combine the hacks and were able to both take arbitrary charge pods offline, fiddle with the power regulation to generate potentially dangerous current draw scenarios, and simultaneously present the operator dashboard with information indicating that everything was just fine. While some of that did require opening a panel and connecting to a USB debug port in one of the pods, with a variety of vehicles parked up and a pre-attack recon of the CCTV setup, I suspect it wouldn't be too hard to engineer things so that one of the pods was hidden from the cameras long enough for you to attach the required cable and replicate the attack in the wild.

Re:

[Smidge204]

Some units including the ChargePoint (which I own one) rely on software to set the upper current limit. During installation you tell the unit what the breaker size is, and while you can use the user's app to change the max current at any time, it will not let you go above the breaker rating set during installation. Changing this requires using a special installer's app and connecting to the unit locally via bluetooth.

So if an attacker gains arbitrary code execution ability, I suppose it might be possible to

Re:

[tlhIngan]

So if an attacker gains arbitrary code execution ability, I suppose it might be possible to set the max current above the breaker/circuit rating.

Assuming that, if a lot of things go wrong in very specific ways there is a chance of starting a house fire by overloading a circuit. A LOT of things would need to line up though, including the use of a plug and socket and/or faulty installation of the circuit, so that risk is incredibly low. I think the worst plausible damage would be constant nuisance tripping of

Re:

[aRTeeNLCH]

Considering the mentioned Alpitronic is a two times 25kWh charger, some of the bad things a hack could allow is to charge for free. That would probably provide a nice incentive to fix the bugs...

Re:

[Vomitgod]

pretty scary...

Stop being professionally offended.

Exploits in the Infotainment system?

[Weirsbaski]

They found bugs in infotainment systems? Wonder if I could use that to fix the damn bugs in the media s/w in my one vehicle (and the enshittifications in the other, newer vehicle)?

Another big-ass vuln...

[sinkskinkshrieks]

Many manufacturers' TPMS systems run unsecured IP that's on the same bus as everything else. Most cars are security Swiss cheese.

Details on what they can do with the DC charger?

[Zarhan]

What can these exploits do?

For the "slow" chargers (EVSEs), I'm not too worried. Worst cases:
- They could allow someone to do "unauthorized" charging if it's behind a key
- They could potentially do a denial of service (make the EVSE not connect)
- If you have limited the current in the charger configuration, the EVSE basically signals the car "you can slurp up to X amperes", if you can hack the EVSE to tell car to grab as much as you can you could, I guess, blow a fuse or in worst case, start a fire. Or just

Re:

[misnohmer]

It is not the worst case scenario. If you push you imagination a little further, imagine the DC chargers become the distribution for malware into the cars. With the "incredibly complicated", as you described it, charging protocol, it's near guaranteed at least some of the EV's can have a chain of exploits that will enable malware to jump from system to system (remember Jeep's OnStar hack about a decade ago?). Now let's say you can flip the logic on the pedestrian avoidance software to turn them into pedestr

Just wait

[ScooterBill]

"Leave the World Behind" - it wasn't supposed to be a documentary.