That earth-shattering NSA crypto-cracking: Have spooks smashed RC4?
Ageing cipher at heart of HTTPS and VPNs fingered by experts
Analysis Fresh revelations from whistleblower Edward Snowden suggest that the NSA can crack TLS/SSL connections, the widespread technology securing HTTPS websites and virtual private networks (VPNs).
Although reports from the New York Times and its allied publications held off on the specifics, it may all mean that US spooks can reliably crack RC4, a popular encryption cipher, security experts fear.
As reported last night, the NSA and GCHQ are able to foil basic safeguards that supposedly ensure privacy on the web, allowing spooks to peek inside the encrypted contents of VPN traffic, online banking and shopping, and more. No wonder Blighty's intelligence agents shrugged off Brits attempting to hide their data in SSL.
The UK and US governments can probably compromise HTTPS connections by gaining access to the root certificates and encryption keys, exploiting backdoors in equipment and algorithms, or otherwise allow the signals intelligence agencies to run man-in-the-middle attacks on encrypted traffic. GCHQ is alleged to have broken the security on some 30 VPN systems, and has plans to get into 300 by 2015.
The NSA's highly classified Bullrun programme allows the agencies to inspect data sniffed from submarine cables, and might involve, at least in part, collaboration with unnamed technology companies.
Tough ciphers, such as 256-bit AES, remain unbroken by the g-men, it's hoped. Snowden himself famously said "encryption works. Properly implemented strong crypto systems are one of the few things that you can rely on".
What the NSA appears to have done is circumvent or nobble the software and hardware that underpin widely used encryption systems, rather than all-out breaking the mathematical foundations of modern-day cryptography.
However, the agency's unspecified "groundbreaking cryptanalytic capabilities" could include a practical attack on RC4.
"Most major SSL-enabled websites use RC4, which was designed in 1987. This NSA crypto story should be a wake up call for the tech industry," said Christopher Soghoian, principal technologist and senior policy analyst at the ACLU in an update to his personal Twitter account.
Encryption guru Bruce Schneier concurred that an attack against RC4 was at least a more than plausible theory. "I don't know one way or the other, but that's a good speculation," Schneier said in blog post.
RC4 is an ageing but still widely used stream cipher that's often used as a component of Transport Layer Security (TLS) and Secure Sockets Layer (SSL) protocols of HTTPS to protect sensitive web traffic from snooping. Security researchers have shown data encrypted by the algorithm can be carefully analysed to silently extract information, such as an authentication cookie used to log into a victim's Gmail account.
This particular attack requires at least 16,777,216 captured sessions but it's easy to imagine that code breakers at the NSA and GCHQ are far further ahead of the game and have come up with a far more elegant, and therefore practical, attack.
RC4 was invented by Ron Rivest in 1987. Various attacks have been developed against RC4, allowing determined hackers to break the encryption, but the technology is still widely used - it's also used in Wi-Fi WEP protection.
It's understood about 50 per cent of all TLS traffic is protected using RC4 and its use is, if anything, growing after the discovery of various attacks (including BEAST and Lucky 13) against Cipher-block Chaining (CBC), a mode of encryption used by TLS.
Finding the internet's skeleton keys
Another theory is that the NSA is exploiting a weakness in crypto key management. This might be a weakness in the way that keys are generated, or it could be a weakness in the way that the secret data is stored.
Dave Anderson, senior director at Voltage Security, commented: "It seems likely that any possible way that the NSA might have bypassed encryption was almost certainly due to a flaw in the key management processes that support the use of encryption, rather than through the cryptography itself. So, is it possible that the NSA can decrypt financial and shopping accounts? Perhaps, but only if the cryptography that was used to protect the sensitive transactions was improperly implemented through faulty, incomplete or invalid key management processes or simple human error."
Voltage markets encryption and key management technology but El Reg's security desk doesn't think that invalidates Anderson's expert perspective, even though there's a degree of self-interest in pointing towards a potential problem close to home. Other experts think the biz may be onto something in warning about possible shortcomings in the way encryption keys are generated.
"Most of the NSA claims could be explained by ability to crack 2048-bit RSA, or server certs generated weakly," said Marsh Ray, a security researcher and software developer at Microsoft-owned two-factor authentication service PhoneFactor.
The latest disclosures fail to detail which algorithms and products the spooks can snoop on. Previous revelations have revealed that the NSA routinely stores encrypted traffic transmitted over Tor for subsequent cryptanalysis.
'Weaken encryption and random number generators'
Matthew Green, a cryptographer and research professor at Johns Hopkins University, was quizzed by the ProPublica news website, which partnered with The Guardian and the New York Times in shedding fresh light on the NSA's anti-crypto operations.
His take on what the NSA might be doing is the most complete we've seen. According to Green, spooks are "working with hardware and software vendors to weaken encryption and random number generators", attacking the encryption used by "the next generation of 4G phones", and establishing a Human Intelligence division to infiltrate the global telecommunications industry.
Green singles out weakening the integrity of SSL as the gravest violation of privacy; the NSA reportedly blows $250m a year working on just that.
Meanwhile, Mikko Hypponen, chief research officer at F-Secure, points to leaked evidence [PDF] that the NSA makes cryptographic modifications to commercial security devices to "make them exploitable".
ISPs, DNS services, and bent certificate authorities (CAs) may all potentially be involved in one one or another, as well as software developers and infrastructure firms. Public Key Infrastructure (PKI) - a building block of modern cryptography - and the whole security model underpinning e-commerce and VPNs suddenly looks a whole lot less secure.
And if the NSA and GCHQ have these abilities then it's no great stretch to imagine their equivalent agents in Russia, China, France, Israel, and other nations, all have similar capabilities or are working towards the same goals.
The NSA-GCHQ's multi-front attack against encryption seems to involve manipulating crypto standards (perhaps including the NIST), stealing encryption keys and perhaps even planting backdoors. But all is not lost and what's needed is perhaps a different perspective on the problem of safeguarding privacy from government dragnet-style surveillance. This is not the cryptopocalypse.
"Strong crypto raises the costs - we must build alternatives to spying with economics as well as mathematics," writes Jacob Appelbaum, a leading Tor developer.