Internet AppSec Remains Abysmal & Requires Sustained Action in 2023

Can we build a defensible Internet? To improve the security of the Internet and the cloud applications it supports in 2023, we need to do better, experts say. Much better.

At the beginning of 2022, companies famously scrambled to hunt down and mitigate a critical vulnerability in a widespread component of many applications: the Log4j library. The following 12 months of Log4Shell woes highlighted that most companies do not know all the software components that make up their Internet-facing applications, do not have processes to regularly check configurations, and fail to find ways to integrate and incentivize security among their developers.

The result? With the post-pandemic increase in remote work, many companies have lost their ability to lock down applications and remote workers and consumers are more vulnerable to cyberattacks from every corner, says Brian Fox, chief technology officer for Sonatype, a software security firm.

“Perimeter defense and legacy behavior worked when you had physical perimeter security — basically everyone was going into an office — but how do you maintain that when you have a workforce that increasingly works from home or a coffee shop?” he says. “You’ve stripped away those protections and defenses.”

As 2022 nears its close, companies continue to struggle against insecure applications, vulnerable software components, and the large attack surface area posed by cloud services.

**The Software Supply Chain’s Gaping Holes Persist

Even though software supply chain attacks grew 633% in 2021, companies still do not have the processes in place to do even simple security checks, such as weeding out known vulnerable dependencies. In March, for example, Sonatype found that 41% of downloaded Log4jcomponents were vulnerable versions.

Meanwhile, companies are increasingly moving infrastructure to the cloud and adopting more Web applications, tripling their use of APIs, with the average company using 15,600 APIs, and traffic to APIs quadrupling in the last year.

This increasingly cloudy infrastructure makes users’ human fallibility the natural attack vector into enterprise infrastructure, says Tony Lauro, director of security technology and strategy at Akamai.

“The unfortunate truth is that no matter what is happening in the enterprise and how well you lock it down and secure it, there is opportunity to attack the users,” he says. “With ransomware and malware, phishing and scams, even if the back end is secure, they can take advantage of the user.”

****Cyberthreats Against Applications Only Loom Larger**

To see an example of how little progress cybersecurity has made in the past three decades, companies do not have to look further than phishing. The social engineering technique has been around for almost as long as email, yet the vast majority of companies (83%) have suffered a successful email-based phishing attack in 2022. Phishing easily leads to credential harvesting and then to compromises of Web applications and cloud infrastructure.

The simple technique can bypass multiple layers of application security and give attackers access to sensitive data, systems, and networks, Daniel Cuthbert, global head of cyber security research at Banco Santander, said at this month’s Black Hat Europe security conference.

“You should be able to click on something and not have it push a reverse shell out to somebody else,” he lamented. “Is it that hard to ask?”

Attackers are also focusing on targeting applications in ways that get by many of the security controls that are operating at the edge of the network.

At the Black Hat Asia conference in May, researchers outlined ways to sneak attacks past web application firewalls (WAFs) to deliver malicious payloads to otherwise-protected applications and their databases. In December, cybersecurity firm Claroty demonstrated more general attacks using JSON to bypass five major WAFs, including those of Amazon Web Services and Cloudflare. In the same month, a pair of researchers used a vulnerable version of Spring Boot to bypass Akamai’s WAF.

Companies have to be more tactical about how they rely on WAFs, says Akamai’s Lauro. So-called “virtual patching” — when the WAF is used to block the exploit of vulnerabilities that are not yet, or cannot yet be, patched — is an important capability. Yet, too many companies use WAFs to protect poorly designed applications, he says.

“You need to identify how that vulnerability could be attacked from the Internet, and virtual patches helps there, but once you are inside the network, the first thing I’m going to do as an attacker is look for some of these zero-days and use them to move laterally,” he says.

Future AppSec Requires Innovation

Efforts to protect the fundamental components of software by securing the software supply chain will be a key source of innovation in the near future. These advances take time to implement and are not silver bullets, but they can result in far more robust software development and end product, experts say.

Providing developers more information about the components they import into their own software through systems like Scorecard, for example, has significant security benefits. Scorecard checks a variety of software project attributes, such as whether there are binary code included in the software, have dangerous development workflows, or has signed releases. Just that information can determine whether a project is vulnerable with 78% accuracy, according to the Open Software Security Foundation (OpenSSF).

Sigstore, which allows each software component to be signed, is another technology that will help developers understand and secure their supply chains, says John Speed Meyers, principal security scientist at Chainguard, a software security firm.

“A key building block for preventing software supply chain compromises is the widespread use of digital signatures,” he says. “This helps reduce the chance of software supply chain compromises and reduce the blast radius when they do happen.”

Companies Can Make Cyber-Secure Application Choices

While those advances in the software development process can result in more secure software, the choice of language can make a significant difference as well. Memory-safe languages can all but eliminate pernicious classes of software flaws, such as buffer overflows and use-after-free vulnerabilities.

Google, for example, found that the use of memory-safe languages, such as Java and Rust, rather than C and C++ resulted in lowering the number of vulnerabilities from 223 to 85 over three years.

Companies need to give developers more support and leeway in selecting secure tools and frameworks, not just focus on productivity and features, says Sonatype’s Fox.

“There is a new reality that companies need to wake up to and deal with, and that is that the developers at the end of the day are the ones that have to make these changes, and the organizations need to recognize their problems and support them,” he says. “Developers are finding their own tools, and they know that’s a problem, but they are not getting the support from the company, so even in a world where developers want to do the right thing, their companies are holding them back.”

At the executive level, companies also need to be using their buying power to focus on holding their vendors accountable for security of their products, Banco Santander’s Cuthbert said during his Black Hat Europe keynote.

“When we look at buying product, and we look at buying software, the reality is that we have zero input to make sure that those vendors, those products are secure,” he said. “We just don’t have that power and we don’t have meaningful influence.”