The rise of AI in cybersecurity 2026: technical risks and defense solutions

The rapid development of artificial intelligence (AI) is no longer a story of the future but has directly reshaped the landscape of the cybersecurity industry in the present. Technology from giants like Anthropic, OpenAI, or Google currently allows hackers to detect computer system vulnerabilities at a vastly superior speed compared to previous decades.

However, similar to any cyber tool, AI is being used simultaneously for both offensive and defensive purposes.

AI reshapes the balance of attack and defense

Reality shows that the time boundaries in cyberattacks are significantly narrowing. According to expert deSouza, if attackers previously could take minutes to infiltrate computer networks without AI, with the assistance of artificial intelligence, breaches now happen in just a few seconds. Hackers are fully utilizing the latest AI programming tools like Anthropic's Claude Code and OpenAI's Codex to create AI agents capable of automating the identification and exploitation of software vulnerabilities.

The explosion of phishing and social engineering using AI

Conversely, software developers and security experts are also racing to use AI itself to patch vulnerabilities. Daniel Stenberg, who runs the important and popular open-source project named Curl, stated: "These AI models are amplifying what humans can do." Thanks to AI, the speed of identifying legitimate bugs has increased to an astonishing level.

Typical security risks triggered by AI

1. The explosion of phishing and social engineering using AI

According to threat-intelligence analysis from Microsoft, threat actors are exploiting AI to create phishing lures, summarize stolen data, and research targets. Notably, the World Economic Forum has also issued warnings about new forms of social engineering combined between realistic emails, deepfake media, and forged documents with extremely high authenticity.

Generative AI completely eliminates grammatical and spelling errors, while being able to create multilingual phishing campaigns. Instead of sending mass emails, cybercriminals now use public data to automatically generate messages directly targeting the job title, company, and habits of each victim.

2. "Code bloat" syndrome and vulnerabilities from open source

The application of AI for code generation has led to a phenomenon known as "code overload" or "code bloat". Open-source projects are currently flooded with code snippets added by AI. Although it helps accelerate programming speed, the flaws in the AI-generated code can directly lead to security vulnerabilities or crash the system. The issue is that determining the responsibility for reviewing these code segments to ensure safety compliance remains very vague.

3. Hazards from deepfakes and content manipulation

Deepfake products are becoming more sophisticated by the day. Jens Kramosch from the organization Leak.Red (Germany) said: "If we talk about the deepfake AI tools that appeared on the net a year ago, since then, at least 23 more advanced models have been created." To cope, many protection software (with reference prices like Leak.Red's at €99, equivalent to RM460.50/month) are having to use AI itself to fight against AI.

Deepfake products are becoming more sophisticated.

According to Nicholas Müller from the Fraunhofer Institute for Applied and Integrated Security (Germany), these software will evaluate the reliability of digital content on a scale from 0 to 100, based on abnormal traces at the pixel level that the human eye cannot see.

4. Cloud misconfigurations and new architectures

Besides the direct factors from AI, security reports also indicate that cloud misconfigurations are currently one of the biggest risks to AI security. This is a fatal weakness that helps hackers manipulate training data, extract sensitive information, and compromise AI models. Furthermore, the rise of serverless architectures and low-code/no-code platforms is also opening up completely new attack surfaces.

Update: The cybersecurity context in Vietnam and the world

In the context of globalization of threats, in early 2026, Vietnam, as well as many countries in the region, recorded a sudden increase in ransomware attack campaigns using AI. APT (Advanced Persistent Threat) groups are automating the reconnaissance phase to find unpatched vulnerabilities (zero-day) in financial institutions and critical infrastructures, posing an immense challenge for security operations centers (SOC) nationwide.

Security solutions in the AI era

(Additional reference information)

The change in attack methods requires a proactive defense strategy and a zero trust system architecture. To neutralize the sophistication of hackers using AI, corporate organizations need to deploy in-depth security ecosystems:

1. Attack surface management & establishing secure AI environments: Requires embedding security policies right from the development and operations phase (DevSecOps). Instantly remediate misconfigurations on the cloud to protect training data.

2. Risk assessment and penetration testing: Need to conduct continuous vulnerability scanning combined with penetration testing to simulate automated attack techniques of AI. From there, classify risks based on actual impact on the business.

3. Enhancing proactive defense (system hardening & EDR): Minimize the attack surface through secure configuration, eliminating unnecessary services. Deploy endpoint detection and response (EDR) solutions on mobile devices to monitor remote workforces or hybrid work models.

4. Professional incident response services: When attacked by malware or automated plugin attacks, an emergency response process that helps isolate, eradicate, and recover the system is a mandatory requirement.

www.ipsip.vn

Over 82% of Vietnamese Enterprises Shift Toward Security Operations Centers (SOC)

Discover why 82–96% of Vietnamese businesses are building or outsourcing a Security Operations Center (SOC). Stay updated on 2026 security trends and optimized solutions.

Businesses can establish multi-layered defense networks and receive consulting to deploy hardening, pentest, and incident response solutions that meet international standards through the network of experts at IPSIP to protect digital assets against the disruptive era of AI.

References:

• AI is on its way to upending cybersecurity (The Star)

• Like analysing a crime scene: Experts explain how to expose deepfakes (The Star)

• The Big Bang: AI has created a code overload (The Star)

• It's Here - How AI is Already Transforming Cybersecurity Today (SEIRIM)

• Top Annual Cybersecurity Reports 2026 (SEIRIM)