The evolution of the cyber business environment continues to become more complex and difficult to secure. Today, businesses are deploying more sophisticated machine learning (ML) models to protect their data and networks from cybercriminals; however, hackers are starting to adopt the same technology to defeat security, identify vulnerabilities and launch attacks.
“We’re not being attacked by human beings anymore,” notes Gordon Gottsegen in an article for Built In, an online resource for tech start-ups. “Computers are attacking us; software is attacking us.”
How Is Machine Learning Used in Cybersecurity?
Machine learning models involve statistical algorithms that identify trends and patterns in existing data to create baseline curves.
With all new data, the technology modifies its algorithms without human input to provide continuously updated, forward-looking intelligence via dashboards and other visualizations. This enables non-technical managers and executives to make decisions more quickly and accurately.
When used to support enterprise information technology security, ML and artificial intelligence (AI) analyze large data sets produced by enterprise networks and identify anomalous trends in the data stream.
“Once the baseline of behavior is built, the AI-algorithms can flag statistically important deviations and alert cybersecurity analysts that further investigation is a must,” according to AI/ML technology provider USM.
What Are Emerging Trends in ML and Cybersecurity?
More businesses are integrating ML and AI into their cybersecurity strategies, and cybercriminals are following along.
“Combining new methods in new ways to create novel approaches … and adversaries show no signs of slowing down,” warns the Darktrace Cyber AI Research Centre, a global cybersecurity consultant based in the U.K.
CSO recently looked at several ways cybercriminals who have the technical skill, computing resources and financing to harness ML will be able to “launch bigger, more complex attacks,” including:
Spam and Phishing Emails: ML has the potential to make a well-known hack strategy into a weapon that can shred enterprise cybersecurity. ML can be used to manipulate security scoring, giving companies a false sense of confidence; deliver phishing emails one-off instead of by bulk delivery, making them more difficult to detect; and generate fake personas to make the fraudulent emails look legitimate.
Password Theft: ML can learn how enterprises control their security — password protocols and periodic updates, for instance — to simplify hackers’ access to networks and data.
Deep Fakes: CSO calls this strategy that generates audio and video counterfeits of real people “the most frightening use of artificial intelligence.” Hackers routinely use deep fakes to produce scam photos, profiles and emails. However, AI takes the tactic to the next level, giving criminals the tools they need to spread their attacks through telecommunication and video technology.
AI Poisoning: Hackers can flood a machine learning model with malicious data to corrupt its output. The IEEE cites a 2016 attack on Tay, a Microsoft chatbot on Twitter. Attackers acting in concert interacted with Tay, feeding it tens of thousands of bigoted and inflammatory messages. Tay’s ML model learned from them, and “started tweeting highly offensive things.”
AI Fuzzing: Cybersecurity experts use this ML technique to discover network vulnerabilities so they can develop patches for them. However, CSO notes that cybercriminals are experimenting with the same strategy to find vulnerabilities to attack before deploying a patch, a process known as zero-day.
While these technologies are critical to preventing cyberattacks, Built In notes that the adoption of AI as a cybersecurity and cyberattack measure is growing, but AI cannot solve every problem. “Humans are still crucial,” it says.
What Makes Advanced Specialization in Cybersecurity Worthwhile?
The career path for cybersecurity-literate business professionals is wide open, according to a survey conducted by the Corporate Governance Institute (CGI), which found:
- Nearly 96% of executives have shifted their cybersecurity strategy due to COVID-19
- Forty percent of executives say they are accelerating digitization
- There are 3.5 million vacant cyber jobs
- Of those polled, 55% of respondents are not confident their cyber spending targets their most significant risks
“The survey found that 51% of executives planned to hire full-time cybersecurity personnel in the next year; 22% said they would increase staffing by at least 5%,” CGI said.
An advanced degree in cybersecurity will equip professionals with the in-demand skills employers and organizations need to protect their systems from cyberattacks.
Learn more about the University of Texas Tyler’s Master of Business Administration with a concentration in Cyber Security online program.