6 AI Cybersecurity Startups Keeping You Safe
The war between machines won’t likely be fought across some bomb-blasted hell-scape, with humans scuttling about like roaches trying to avoid being squashed. Rather, machines will fight it out over fiber optic connections, with the battleground being computer servers and laptops containing valuable information. You’ll recall that monochromatic pantsuits weren’t Hilary Clinton’s only problem: Russia (or some obese, Big Gulp-slurping teen in his mom’s basement) hacked her private emails. Cybersecurity is still the domain of humans, but the job is increasingly being turned over to predictive systems that use various forms of artificial intelligence that do everything from protecting financial information to detecting fraudulent behavior.
It’s no secret that cybersecurity is big business. Forbes reported that the sector is projected to reach $170 billion by 2020. Computer World UK, quoting analyst firm ABI Research, estimated that AI cybersecurity will mean big spending on big data, intelligence, and analytics, to the tune of $96 billion by 2021. That growth is certainly being reflected in the startup landscape. Analytics firm CB Insights found that in 2016 there were more than 400 deals to cybersecurity companies, up from 378 in 2015. The category has been trending up significantly since 2012. In 2015, funding hit nearly $4 billion, according to CB Insights, and last year reached about $3.5 billion in investments.
During January’s Innovation Summit, CB Insights revealed its top AI 100 companies. Among the best of the best were six startups playing in different sectors of the cybersecurity game. Let’s take a look at what makes these AI cybersecurity companies stand out from the competition.
Sporting the tagline “stopping tomorrow’s attacks today,” Cylance is the only unicorn (startup valued for at least $1 billion) in this AI cybersecurity group. It was only one of five of the AI 100 that had reached such a valuation as of January. To date, it has raised $177 million, including a $100 million Series D mega-round last year, led by Blackstone Tactical Opportunities (not the same Blackstone that created super assassin Jason Bourne, we assume). Among its many other backers is Khosla Ventures, which has participated in four investment rounds so far, and is one of the top VC firms for AI startups, according to CB Insights.
We first profiled Cylance last year as one of nine hot cybersecurity startups. We marveled at its ability to use machine learning to protect computers from attacks using just 1 percent of CPU and 60MB of memory on a machine without any internet connectivity. Its flagship product, CylancePROTECT, is antivirus for malware that doesn’t even exist yet; it uses a mathematical approach to detect threats before they begin—like a good wingman at a bachelor party. The company boasts 6,000 customers worldwide and claims triple-digit growth (+322% in 2015 and +607% in 2016) over the last two years.
So those numbers made us curious to read a report published just this month in The Register that the Irvine-based startup was laying off up to 20 percent of its workforce. Cylance did confirm to the web publication that it was undergoing a restructuring but did not confirm any job losses. The company did issue a statement, according to The Register, with the usual corporate mumbo jumbo about realigning resources, but it also claimed that it was on pace to more than double revenue in 2017. The Register story also claimed that Cylance’s software is racking up false positives—flagging ordinary computer operations as malicious—that may be hurting its business. However, the company reports an efficacy rate of more than 99 percent against malware and a false positive rate of .000314 percent.
Cambridge-based Darktrace sounds similar to Cylance in terms of its AI cybersecurity system, detecting and responding to previously unidentified threats using machine learning and mathematics. Founded a year after Cylance in 2013, Darktrace has raised a respectable $104.5 million, including a $64 million Series C last July. Claiming 400 employees across 24 global locations, Darktrace recently announced it had secured more than $150 million in contracts in its last fiscal quarter and boasted a 500 percent increase in annual growth for contracts in the Asia Pacific region for its Enterprise Immune System.
Its platform uses AI algorithms to detect threats in different industries and on various types of networks, including physical, cloud and virtualized networks, as well as Internet of Things and industrial control systems. One case study of a multi-billion-dollar company found that the client was able to recuperate the cost of Darktrace’s Enterprise Immune System by eliminating other antivirus systems, as well as reduce mean detection time of threats by 40 percent. We took a much closer look at this fast-growing startup in an article titled “Darktrace Creates Enterprise Immune System Using AI“.
Deep Instinct, based in Tel Aviv, claims to be the first company to apply deep learning to cybersecurity, despite being founded in 2014. No details are readily available on its funding, but backers include Blumberg Capital and U.S. News & World Report. Its stat sheet should sound familiar to us by now: Deep Instinct’s AI cybersecurity system can anticipate a threat against previously undefined intrusions across devices and servers. That protection extends to endpoints, such as a laptop, even when the device is not connected to the internet.
SparkCognition does a little bit of everything, recently adding AI cybersecurity to its machine-learning platform. The Austin, Texas startup has grabbed more than $6.3 million in investments. The most recent—apparently 2016 was a very good year for cybersecurity deals—coming last April on a $6 million Series B. SparkCognition calls its AI cybersecurity platform DeepArmor. In addition to leveraging neural networks (another way to say its computer uses complex algorithms based on how the human brain works), DeepArmor also employs natural language processing to understand and research potential threats. A second product called SparkSecure also uses NLP to help human security analysts do their job more efficiently. You can learn more about DeepArmor directly from SparkCognition’s nameless AI personality in the following video:
Update 10/08/19: SparkCognition has raised $100 million in Series C funding to invest in growing its research group and continue expanding globally (it recently set up operations in Latin America and in the Middle East). This brings the company’s total funding to $163.2 million to date.
Now we’ll shift away from the AI cybersecurity version of antivirus software to machine learning that detects fraudulent behavior. Sift Science has raised $53.6 million over four rounds. A Series C last July, led by Insight Venture Partners, added $30 million to the cyber coffers. Its software platform automatically learns and detects fraudulent behavioral patterns, then alerts businesses before they or their customers are cheated. Its cloud-based platform is backed by more than 16,000 fraud signals that are updated across a growing global network of more than 6,000 websites and apps. Like other forms of machine learning, the more data you feed Sift Science, the smarter its AI performs against fraud. Monthly plans range from $500 to $10,000 depending on the size of the company and its exposure. Clients include well-known names such as Twitter, Airbnb, and Zillow.
Shift Technology applies its AI cybersecurity solution to detect fraud in the insurance industry. The Paris-based startup has collected $11.8 million in two rounds, including a $10 million Series A in May 2016. Accel Partners, one of our top 12 VC firms, led the funding round. The FBI estimates that insurance fraud outside of health care costs the U.S. economy alone more than $40 billion per year—and that’s just Trump Tower in New York. There’s obviously quite a bit of incentive to cut down on fraudulent claims. Shift Technology, which takes a Software-as-a-Service approach, has analyzed more than 75 million insurance claims to date. TechCrunch reported that Shift’s AI algorithms catch the bad guys about 75 percent of the time. Each claim is tagged with the potential risk of fraud and provides insights—the “why” and “how”—on what makes the claim suspect.
Some cybersecurity experts warn that AI cybersecurity is still too new to be considered reliable as a wholesale solution. They complain that machine-learning platforms return too many false positives or that human hackers still hold an advantage over AI agents. Of course, some of those same experts have skin in the game. The predictive power and analytic capabilities of an AI cybersecurity system will only get better—and so will the threats they must detect. The winner in this particular arms race will be the startup that can build the best brain for the job.
If you enjoyed this article, then sign up for our free newsletter - Nanalyze Weekly. About every week, we'll send you a simple summary of all our new articles. If you didn't enjoy this article, share it on Twitter and tell everyone how much you hated it.