+− THE DAILY DIFFdev & AI news
NEEDS REVIEW

Connected-car identity security: owner authority and physical proof

Shubham Shah and Sam Curry report administrative identity access that led to owner-equivalent connected-car app authority on permissioned vehicles.

Want to see our stories first on Google? Add us to your preferred sources — it helps us grow.

Shubham Shah and Sam Curry report administrative identity access that led to owner-equivalent connected-car app authority on permissioned vehicles. Their November 7 publications describe earlier research. The US companion reports same-day remediation; the separate Australian July account leaves its current remediation status unspecified.

Read the written edition (English) ↗

What this video covers

  • Administrative control of trusted customer identity can affect the vehicle authority granted by downstream app services.
  • Curry limits complete vehicle-chain testing to US accounts and vehicles; regional directory reach and Shah's broader combined claim have different scopes.
  • The approximately two-hour response belongs to the US companion's disclosure timeline. The Australian account gives no matching remediation date.
  • Supported commands depend on model, subscription and active connected services. Driving or immobilizer defeat was not established.
  • The researchers refreshed vehicle status and confirmed physical effects with the owner beside the vehicle.
  • PingIDM's versioned production guide recommends removing or protecting development tools and identifies a separate console authentication system.

Questions and answers

How did identity-server access become connected-car authority?

Administrative identity access let the researchers obtain owner-equivalent app authority, according to Shubham Shah and Sam Curry. The connected service associated customer accounts with vehicles and supported commands. Their account distinguishes administrator access, a completed customer session and the mobile app tokens accepted by its services. Demonstrations used permissioned vehicles.

Watch this part 0:00 · Shubham Shah and Sam Curry · Shubham Shah

What do the reported dates and regional scopes establish?

The US companion describes earlier research and a same-day fix after disclosure. Curry limits complete vehicle-chain testing to US accounts and vehicles, while affected identity directories served multiple regions. Shah separately describes Australian research beginning in July. His article leaves the current Australian remediation status unspecified, so the sources do not establish one worldwide discovery or repair timeline.

Watch this part 0:33 · Shubham Shah and Sam Curry · Shubham Shah

What does production hardening guidance say about development consoles?

Remove or protect development and debug tools before production, according to Ping Identity's versioned security guide. It identifies the console as having authentication separate from the identity product's authentication. That distinction makes a console its own review boundary. The episode recommends inventorying administrative surfaces and checking which authentication system protects each one.

Watch this part 1:41 · Ping Identity · Shubham Shah and Sam Curry

How did the researchers confirm a physical vehicle result?

They refreshed vehicle status and checked the physical result with the owner beside the vehicle, according to Shah and Curry's US companion. API acceptance alone did not prove action. Shah unlocked and started the vehicle, then stopped it and locked it again. The reported evidence covers supported app commands; driving away or defeating an immobilizer was not established.

Watch this part 4:48 · Shubham Shah and Sam Curry

What was The Daily Diff's verdict on the identity trust boundary?

The Daily Diff's verdict was NEEDS REVIEW. Niko would audit administrative paths into trusted customer identity before relying on them for physical controls, while crediting the reported US fix. Shah and Curry's accounts show why upstream control over identity records matters to downstream authorization. The verdict does not assert an ongoing worldwide vulnerability.

Watch this part 5:10 · Shubham Shah and Sam Curry · Shubham Shah · Ping Identity

Transcript

How did a login become a car key?

0:00 Your car's app checks that you're the owner. Researchers made its identity server vouch for them instead. In this video, how did login become vehicle authority? And what proved a car actually obeyed? It's Monday, November ninth, and this is The Daily Diff. One final check separates a convincing screen from an unlocked car. Connected car services are the online systems behind an owner's phone app. Shubham Shah and Sam Curry report that administrative identity access let them

0:25 obtain the same app authority as an owner. They demonstrated remote commands on permissioned vehicles. Their US account says the flaws were fixed after disclosure.

What is newly published, and what was already fixed?

0:33 Both writeups appeared on November seventh. Curry dates his disclosure to early this year. Shah describes Australian research beginning in July. Those are different timelines. Reading them as one worldwide emergency would give the headline more horsepower than the evidence. Shah bought one of the unnamed automaker's SUVs in twenty twenty-three. Which internet service receives the tap when I unlock my car?

0:53 He and Curry spent two weeks examining the identity system. The key fob has acquired a backend team. The app signs you in and loads your garage.

Which identity does the vehicle service trust?

1:01 The connected service associates that account with vehicles and permitted commands. That relationship is the useful answer. Whoever controls the trusted customer identity can inherit capabilities intended for the customer. In Australia, a public management tool exposed customer information through interfaces protected by keys shipped in browser JavaScript. Ambitious definition of secret. Shah also found identity administrator credentials in a vendor bundle,

1:24 and reported that they worked in production. The perimeter blocked sensitive routes. But software layers interpreted the same request differently. Curry says the edge treated a path as static content while the backend reached an administration console. The welcome mat and the locked door were consulting different maps.

Why did a development console matter?

1:41 That console still accepted default administrator credentials, according to Curry. The production guide already tells operators to remove or protect it, and says its authentication is separate from the identity product's authentication. Another login hiding between the checklist's bullet points. The exposed administration tooling allowed code to run inside the identity service. The researchers accessed customer records behind the public filtering boundary. A perimeter filter checks traffic crossing that perimeter.

2:06 A process already inside the service has a different relationship with the data. This explains the multi-factor problem in the US account. Identity records supplied the login information and destination for an email verification code. Administrative control over those records let the researchers complete the login. The upstream definition of the customer had already changed. Shah describes a separate Australian authentication misconfiguration that let his chain skip the extra factor.

2:31 Calling multi-factor authentication broken everywhere would erase the configuration details developers need. Keep your extra factor. Keep the system defining its owner trustworthy too. The credentials have different jobs.

When does a session become app authority?

2:43 Administrator access controls identity records. A customer session represents a completed sign-in. The mobile app obtains tokens accepted by its services. Calling all of those a magic token loses the boundaries that made owner-equivalent access possible. Curry describes two unnamed automakers' identity directories across multiple regions. He limits complete vehicle testing to US accounts and vehicles. Shah describes one automaker across two Australian brands,

3:09 then claims broader combined reach. Those different scopes give us no basis to identify the companies. Hacker News argues about whether phone unlocking should exist. One reply points to checking a forgotten lock and cooling or warming the car. People like useful features. Giving customers convenience also gives the service an expensive job. It has to earn its authority.

What did remediation actually cover?

3:29 Curry reports both automakers fixed the vulnerability about two hours after receiving it and rotated affected credentials. That is the US companion's timeline. Shah's July account supplies no matching remediation timeline. His account leaves Australia's current remediation status unspecified. We can't turn the two-hour response into a global repair stopwatch. For developers, start with the administration surface. Inventory what is exposed and which authentication system protects it.

3:55 Remove unused development tools. Protect the tools you retain. Review each layer's request interpretation before treating an edge rule as the last word on access. Shah says the Australian chain finally worked around midnight, opening a car they were allowed to test. Identity plumbing with a satisfying ending. A writeup earns its conclusion by checking the other end.

4:14 The backend's confidence is only half the conversation. Curry's US account says capabilities depend on the vehicle model, subscription, and active connected services. Supported commands include door unlock and remote engine start. Neither writeup establishes driving away or defeating an immobilizer. The evidence covers particular app capabilities and permissioned physical demonstrations. Ask which service can change the facts your authorization relies on. Password resets and contact changes matter because other systems consume

4:41 them as truth. A valid login can rest on a rewritten identity. This car story becomes your application story there.

What proved the car actually obeyed?

4:48 The final check happened beside the car. Curry says an accepted API command alone proved nothing about the car's behavior. They refreshed its status and confirmed the physical result with the owner standing beside it. Shah unlocked and started the vehicle, then stopped it and locked it again. The proof reached the driveway. If you'd rather read this than hear me say it, the diff lands in your inbox every morning, free at the daily diff dot dev, link below.

Which boundary needs review?

5:10 So, today's verdict, needs review. I'd audit administrative paths into customer identity before trusting it with physical controls. The reported US fix deserves credit. Who can redefine the owner that every downstream service believes? And that's the diff for today. I'm Niko from Axrisi. Merge responsibly.

Sources

  1. Hacking Two Automakers: ForgeRock Path Confusion, RCE, and Remote Vehicle AccessShubham Shah and Sam Curry
  2. unlocking any car from X brand in most of the worldShubham Shah
  3. Remove or protect development and debug tools — PingIDM 7.2Ping Identity
  4. Connected-car disclosure discussionHacker News
  5. Sam Curry portrait by Magistr8; cropped/background-masked/resized without generated detail; photograph date unspecifiedWikimedia Commons
  6. Portrait licence — CC0; no endorsement or current-reaction claimCreative Commons
  7. Sam Curry's official About page corroborates portrait identitySam Curry
  8. Jurassic Park fictional-character GIF; silent maximum two-second transformed process commentary, no actual automaker reaction or blanket reuse licenceScreenJunkies / GIPHY
  9. GIF uploader's native Source link; publisher provenance is not a blanket reuse licenceScreenJunkies
  10. Is This A Pigeon template; maximum two-second transformed fictional process satire, no actual automaker reaction or blanket reuse licenceImgflip

Niko from Axrisi

Nikoloz Turazashvili · Vexrail · Axrisi

LinkedIn · dev.to · GitHub · X · Substack

Related videos