Digital certificates are dead - killed by issues of scalability and trust
Digital certificate authorities are neither authorities, nor entirely secure, argues Miracl CEO Brian Spector. Something new is needed
Digital certificates used to be synonymous with security on the internet - a seemingly trusted enterprise issued the certificates that could verify all of our relationships online.
But in recent years, the industry has been plagued by scandals, with hackers accessing legitimate code-signing certificates and bypassing security checks around the world.
While there have been various notable attempts to reform the industry, such as the push for ubiquitous encryption and transparency championed by Let's Encrypt, several big issues remain.
While certificates once successfully authenticated servers, back when everyone was using a single device, they don't scale for the world we live in now, where the number of devices has boomed and their usage has positively exploded.
Certificates can't authenticate users across apps and mobiles, they don't work with virtualised computing, and they certainly won't scale for the Internet of Things (IoT).
So the only question remaining is, what does the future of security on the web now look like?
Authority undermined
The security shortcomings within digital certificates are systemic and architectural. Each certificate authority effectively represents a single point of compromise for hackers to target because they store a single root or master key, which is used to authenticate every single website it secures.
This is a huge vulnerability and means that if a hacker gains access to this key, they can fraudulently issue certificates from that authority. This is like a criminal posing as a police officer by using a real police officer's badge, because there is no way to tell the difference between a fraudulently issued certificate and a real one.
This is why, when Dutch certificate authority DigiNotar was hacked in 2011 and more than 500 fake certificates issued, major web browser makers reacted by blacklisting all of DigiNotar's certificates.
In addition, some certificate authorities generate a key pair for each certificate they issue, meaning that they also know (usually only for a short time) the private key details for each website they encrypt. If a hacker were to gain access to this information, they could potentially decrypt any of the web traffic protected by that authority.
Unfortunately, the vulnerabilities in public key infrastructure (PKI) - the architecture behind the certificate industry - have been common knowledge for at least 15 years. Each hack only exacerbates the problem, and reduces the amount of trust that internet users have in the whole system.
When PKI was first established, all the organisations with public keys were kept in a directory listing their name, address and public key. Like an old-fashioned address book for the internet, this system might have worked in 1976, but it is simply no longer viable for the billion websites in use today.
This problem of scale is probably the biggest issue facing the digital certificate industry today, particularly in light of the huge volume of connected devices involved in the Internet of Things.
Gartner forecasts that the number of IoT devices will grow to 26 billion units by 2020 - a 30-fold increase from 2009. PKI is just one of the technologies in use today that simply isn't ready for the massive scale and highly dynamic nature of the future Internet of Things, and new communication patterns it creates.
The frequency and severity of hacks affecting digital certificates mean that the tide of public opinion is turning against them. One of the reasons that internet users are beginning to mistrust certificate authorities is that they are mostly private companies who operate on a for-profit model.
Yet the language they adopt - by calling themselves "trusted" and "authorities", for example - implies a degree of authority, as if they have somehow been selected by the public to police the internet. But in truth, the major certificate authorities are private, US-based companies who turn significant profits from issuing digital certificates to all and sundry.
This lack of transparency, and the tendency to offer security only to those who can afford to pay for it, were some of the factors that drove the creation of Let's Encrypt by the Electronic Frontier Foundation, the Mozilla Foundation and the University of Michigan.
Distributed trust
Much like the underlying internet protocols themselves, Let's Encrypt is a joint effort to benefit the community, and enables anyone who owns a domain name to obtain a trusted certificate at zero cost. This ought to be a huge step forward and should help to highlight some of the failures of the current system, but unfortunately it doesn't solve the inherent security flaws.
What's needed is to start all over and replace certificate authorities with something completely new. One alternative, which could make the whole internet safer, is the concept of "distributed trust".
Rather than allowing a single certificate authority complete access to a root key - and therefore creating a single point of compromise - the functions of root keys can be split into three parts and distributed across three separate entities. This would mean that the only audience with complete knowledge of the key would be the end user themselves.
Using this system, even if a hacker managed to gain access to one of the distributed trust authorities, they could not gain access to a complete key or be able to issue shares of new ones. This would make current root key compromises and key escrow threats an order of magnitude more difficult since an attacker would need to subvert all three independent parties.
By distributing trust between several locations, rogue people and practices can be self-governed, and the web can continue to grow and expand more securely.
Brian Spector is CEO of identity integrity as a service company Miracl. He can be contacted via the company's website.