Safeguard
Vulnerability Analysis•

CVE-2023-32157: High Heap Overflow in Tesla Model 3 Bluetooth

CVE-2023-32157 is a high (CVSS 7.5) heap overflow in Tesla Model 3 bsa_server Bluetooth handling. ZDI says firmware 2023.20 fixes it. Update.

Safeguard Security Team
4 min read

CVE-2023-32157 is a high severity heap-based buffer overflow, scored 7.5, in the bsa_server process of Tesla Model 3 firmware. An attacker who can pair a malicious Bluetooth device could run code in a sandboxed process. ZDI says firmware 2023.20 fixes it, so update.

What's affected

The NVD record, dated 2024-05-03, covers Tesla Model 3 firmware, listing release 2023.6 in its product data. The flaw is in the bsa_server process: the length of user-supplied data is not validated before it is copied into a fixed-length heap-based buffer, which allows code execution as an unprivileged user in a sandboxed process. The weakness types are CWE-122 (heap-based buffer overflow), CWE-787 (out-of-bounds write). The record vector is CVSS:3.1/AV:A/AC:H/PR:N/UI:N/S:U/C:H/I:H/A:H, scored 7.5.

Zero Day Initiative advisory ZDI-23-973 (ZDI-CAN-20737) says the issue is fixed in the 2023.20 firmware release. It was reported to the vendor on 2023-03-30 and published on 2023-07-18 after coordinated disclosure.

The advisory says an attacker must first be able to pair a malicious Bluetooth device with the target system, which is a physical proximity requirement.

The sources differ on severity. The NVD record scores it 7.5 (High), AV:A/AC:H/PR:N/UI:N/S:U/C:H/I:H/A:H. ZDI scores it 4.6 (Medium), AV:A/AC:H/PR:L/UI:N/S:U/C:L/I:L/A:L, with low privileges required and low impact, reflecting that the code runs in a sandbox. This post follows the NVD record, High, and flags the ZDI figure.

No source says this flaw is exploited in the wild, and it is not in the CISA Known Exploited Vulnerabilities catalog.

How to check if you're exposed

Tesla delivers vehicle fixes as firmware releases. Check the software version shown on the vehicle's touchscreen, in the software information area of the controls menu, and compare it with the release named above.

If the car is not on a fixed release, make sure it is connected to a trusted Wi-Fi network or cellular service so it can receive updates, and install any pending update when the car prompts you. The advisory describes the flaw as requiring a prior foothold, so a fixed release matters most for cars that connect to untrusted Wi-Fi or pair with unknown Bluetooth devices.

The sources give no workaround besides updating.

What to do now

  1. Confirm each Model 3 you manage is on firmware 2023.20 or later. Fleet owners should record the version for every vehicle.
  2. Install any pending vehicle update promptly rather than postponing it.
  3. Avoid joining unknown Wi-Fi networks and avoid pairing unfamiliar Bluetooth devices, since the advisory describes attacks that begin from those paths.
  4. Treat the vehicle as a connected computer in your asset list, with an owner and an update routine.
  5. If a vehicle cannot update, contact Tesla service.

Why it matters

CWE-122 is a heap overflow, reached here through a Bluetooth profile handler. The attacker needs to be near the car and to get a device paired, and the resulting code runs as an unprivileged user in a sandboxed process. That limit explains the lower ZDI rating. It is still a way to run code on the car from nearby, and it is the sort of flaw an attacker could pair with a sandbox escape. Note that the fixed release here, 2023.20, is later than the 2023.12 release that fixes the other two Tesla issues in this set. Safeguard offers a free in-browser dependency scan at gold.safeguard.sh/scan for the packages in your own code. It does not inspect vehicle firmware or installed desktop applications, so those need a separate inventory.

Try it yourself

Click through it at your own pace, or open it full screen.

Quick check

Why is the fixed release 2023.20 and not 2023.12?

ZDI names 2023.20 for this flaw and 2023.12 for CVE-2023-32155 and CVE-2023-32156. Being on 2023.20 or later covers all three.

How serious is the ZDI 4.6 score?

ZDI rates it Medium because the code runs in a sandbox with low privileges. NVD rates it High. The difference is the impact assumed.

Sources

Last reviewed: 19 October 2026.

Never miss an update

Weekly insights on software supply chain security, delivered to your inbox.

We email you a link to confirm, and only then add you to the list. Unsubscribe in one click. Privacy

Self-healing security runs on Safeguard.

Your first fix PR is minutes away.

No sales call required, even your agent can complete the purchase over MCP.