CC ihf who may add another evidence.

I recently got phantom touches on Samus (stable channel) maybe once a week, but a reboot usually fixes them. Didn't file feedback though.

alemate@ and ihf@ which versions of OS & firmware are you guys using?

I am using ToT build.
Yes, device reboot helps for some time. Browser restart does not.
==== bios_info ===
vendor               | coreboot
version              | Google_Samus.6300.174.0
release_date         | 04/02/2015
size                 | 8192 KB
ro bios version      | Google_Samus.6300.102.0

==== platform_info ====
vendor               | GOOGLE
name                 | Samus
version              | MP.B
family               | 

CHROMEOS_FIRMWARE_VERSION     Google_Samus.6300.174.0
CHROMEOS_RELEASE_DESCRIPTION  9202.54.0 (Official Build) beta-channel samus

It seems that it causes weird failures across the OS in general on this firmware.
For example my TPM decided to nuke my home directory and when clicking Alt+Shift+I for feedback, my machine restarted and lost the OS.

apronin@ Could you help us figure this out?

1) Sure, will try to help with the TPM issues. Does it come back after rebooting, or does still got back to the "OS is damaged or missing" screen after reboot? If the latter, could you please press Tab on the "OS missing" screen and attach the screenshot?
2) I'd suggest branching a separate TPM bug and track it there. Let's keep this one focused on random touch events?

If it's consistently "OS damaged or missing" screen, it's probably a "hard" variant of dictionary attack lockout.
If so, it should be recoverable by power-cycling the device about 25 times, each time leaving the device on for at least 20 seconds.

assuming the preliminary diagnosis from comment #8 is right:
To get into this state several factors are needed (I can provide more details separately, but here is a similar bug from the past: b/35578636):
1) the device should be reset at a "bad" moment. This random reset during Alt+Shift+I could have been caused by something not related to TPM at all, but if it happened at the wrong moemnt it could trigger this chain of events.
2) On the next boot, tpm is woken up for a short period of time before going into reset again.

If that's happening, it can not be prevented in the OS. But something similar to this fix in the firmware may help to drastically reduce the frequency of such cases: https://chromium-review.googlesource.com/c/401302/ (from b/35578636#comment118).

Just realized that https://chromium-review.googlesource.com/c/401302/ didn't help much in the other case.
Still, this TPM issue is worth taking another look at, but is probably unrelated to random touch events, unless there's some electrical glitch that leads to both.

It's also possible that there are some issues with the bus. Do the touchpad and the tpm sit on i2c on this board? Anything i2c-related in the logs, if so (once the board is recovered from the tpm error, and the issue is repro'd)?

Thank you all, we couldn't fix it and have returned the device to Chromestop.