I think the Snap has already been deleted. It would be a good idea to introduce a warning message for more careless users. For exaple: āThis Snap is not verified. Donāt share personal data unless you are sure of its sourceā.
Evil actors are using multiple techniques to evade abuse detection, which is unfortunately not easy to be dealt with:
Response from Gemini LLM:
Is this string wierd? āExodÕ½ź±ā
That string, āExodÕ½ź±,ā definitely looks a little weird!
Itās an example of a homoglyph attack or IDN homograph attack, where one or more letters are replaced with similar-looking characters from a different alphabet (like Cyrillic, Greek, or another Unicode block).
Hereās the breakdown of the characters in the string:
E, x, o, d: These are standard Latin alphabet letters.
Õ½: This is the Armenian Small Letter Youn (U+057D). It looks almost identical to the Latin lowercase āuā.
ź±: This is the Latin Letter Small Capital S (U+A731). It looks very similar to the standard lowercase āsā, but itās a small-cap letter.
The intent is often to make the word look like āExodusā while using entirely different characters, which can be used for deceptive purposes in URLs, usernames, or other security contexts.
Itās published again with a bit different name. But anyways itās easily found in App Center by plain ASCII āexodusā query. Maybe because this word is used in the description. Probably mix of different alphabets should be prohibitied? Or at least it should be marked and verified by human?
In general, I think itās not a problem but an interesting case and great opportunity to make the repository safe. It would be really huge competitive advantage if Snap was malware-free! Iāve lost all my money by that malware and I just want that it was not in vain, that it will help others.
First of all, thank you for bringing this to our attention and apologies for the delayed response.
Weāve now fully processed the reported malicious snaps and have removed them from the Store. Weāve also taken additional steps to review our monitoring processes and to ensure that similar reports are caught and addressed much sooner in the future.
Regarding the concerns about our publishing process; we recognize the points made in the original post. Over the past year, weāve been actively tightening [1] our publishing and review policies [2], which in many cases includes manually reviewing the registration requests, to reduce the likelihood of problematic content appearing. While this has already helped lower the number of cases, itās an ongoing effort and we know thereās more to do.
We werenāt monitoring this channel as closely as others, which contributed to the late response but weāve adjusted that now so similar reports wonāt be overlooked. In addition, weāre not introducing formal SLAs at this point, but improving responsiveness remains a priority. Weāre definitely continuing to strengthen our publishing and review processes, and that work is ongoing.
Iām afraid this will just keep on going. Could we consider a policy of disallowing publishing of crypto wallet applications unless the publisher had been verified?
Isnāt that process already in place ? As I understand it the problem here are the āimpostorsā ⦠i.e. packages that pretend to be something else and then ship crypto fishing bits inside ā¦
Clearly the current policy appears to be ineffective. Maybe for once thereās actually something useful that could be done with LLMs here. If Gemini is able to find an issue then maybe thatās how we could approach that. I would not mind if more apps end up needing manual review, but itās better to have more false positives for the is-a-crypto-scam-wallet-app check than make them available to the user.
Suggestion: make a Security category specifically for the reporting and discussion of malware found in the Store? That way, any interested parties can subscribe to it, including policy-reviews et al.
Weāre considering options like that. We realised that reports of this kind were appearing in this category, so weāve begun monitoring it more closely. Our hope is that quicker responses here will encourage more reports, and if the volume grows, it may make sense to create a dedicated category for these cases.
We understand the concern, but itās worth noting that our current checks already filter out a significant amount of malicious or non-compliant content before it ever reaches the store. Thereās still work to be done so weāre always looking for ways to improve the system, but like Lin-Buo-Ren mentioned - itās whack-a-mole.