I never thought I will sign up here because of sad reasons. But shit happens.
After what happened to PicoCrypt I am trying to find new alternative that can encrypt files.
Exactly files. I don’t interested in disk encryption etc as I know everything about LUKS.
I need tool to encrypt file put it on flash (or via insecure channels like WhatsApp or Gmail or whatever) and give it to another person wich should be able to decrypt it.
Any good TRUSTWORTHY solutions?
I am operating on Linux only. Shitdows support NOT needed.
What happened to PicoCrypt: https://github.com/Picocrypt/Picocrypt/issues/134 (sorry for such link format, protections annoying me)
The age crowd likes to hate on gpg, but as best as I can tell they avoid the biggest pitfalls of gpg (difficult UX around key management) by leaving it entirely up to the user to manage key trust which is picking a different pitfall. I don’t know if that results in better or worse real world security outcomes.
Age does have some more modern cryptography choices compared to gpg, but neither is cutting edge enough to address PQC concerns.
I like that age (Go) and rage (Rust) are in memory safe languages, rather than C, but gpg code will have received far more scrutiny over the years.
+1 AES-256 being fine.
I think yubikey resident gpg keys are a reasonable approach for security concerned commercial ventures.
Good to meet PicoCrypt but it’s already dead. However, as the author said it’s secure and fine for continued usage as long as no new security issues are found.
Alternatives ?
If you’re simply using passphrases in PicoCrypt, then all the suggested solutions in this thread are sufficient IMO. Go with one having a nice GUI wrapper that you like.
Perhaps it’s time to abandon passphrase and consider PKI-based solution. Then age/rage are super convenient because they can re-use SSH keys which almost all Linux users possess.
There is mounting evidence of hardware issues with AES encryption.
Like all things security its about what you consider to be your attack surface. If i was going out of my way to encrypt individual files, AES would not be my first choice today.
I don’t see anything that would quality as “evidence” that AES is compromised. The top of my search result reads “The claim of an AES backdoor in Intel processors is largely unfounded. While concerns about backdoors in cryptographic systems exist, particularly in the context of government surveillance programs, the specific assertion of an AES backdoor in Intel processors is not supported by credible evidence.”
The first page of search results that I get supports this. If there were a hardware AES exploit it would be big news that would dominate search results. The rumors of a hardware AES exploit go back years and it looks like there’s still no exploit so rather than “mounting” I’d say the rumor that hardware AES was exploitable has failed to materialize.
Yeah, so far as I know the general route’s a quantum asymmetric crack to extract negotiated symmetric keys, which is still sitting at around ~70 bits for a few thousand €/US$. So only ~1980 bits from breaking RSA-2048.
More realistically, stuff like Downfall (CVE-2022-40982) and Inception (CVE-2023-20569)‘ll continue to be possible routes to key pwnage and the Youtube/Reddit conspiracy theory crowd’ll continue to get search indexed misrepresenting them as hardware backdoors installed by the United States’ National Security Agency or [other favorite cabal]. The fuse encryption key in Intel SoCs is another popular target and gets broken every few years, IIRC usually with physical access requirements. China’s observations Intel has problems are available for tinfoil hat reinforcement but don’t specifically claim AES vulnerabilities that I’m aware oft and aren’t much different from the United States’ disgruntlement with Microsoft for putting profits ahead of security.
Realistically, at what point are you confident enough of symmetric key protections AES becomes the weakest point? And if you actually manage to get there what algorithm would you use besides AES? ¯\_(ツ)_/¯
This would be a huge news story, if there was any truth to it.
There seems to be one C++ programmer (not a security researcher or cryptographer) on Youtube with a few thousand subscribers, who made a video claiming Intel AES is backdoored. Why? Because some output of his software implementation doesn’t exactly match Intel’s, and he can’t believe he made any mistakes in his code. The top comment on his video? “Endianness issue.” That’s it.
I haven’t watched either of his 2hr videos, but the transcripts don’t mention NIST’s AES test vectors, which is how someone who knows what they are doing actually verifies an AES implementation.
You’re free to shut-off hardware acceleration if you want to. You can just disable AES-NI with, e.g.: $ export OPENSSL_ia32cap="~0x200000200000000"
