'AI Creates Threats and Defenses'... What is Google's Cyber Security Solution? [Cloud+]
"Hey Gemini (Google's AI model), check if this code is malicious."
With the advancement of artificial intelligence (AI), cybersecurity threats are becoming more sophisticated, but at the same time, AI is evolving into a 'shield' that detects and responds to them in real time.
Evolving cybersecurity threats with the advancement of AI and cloud
Mark Johnston, Google Cloud APAC's General Manager for Security, Networking and Collaboration, demonstrated a future security system in which AI detects malware with a simple command in his keynote speech at the '31st Information and Communications Network Information Security Conference (NetSec-KR 2025)' held at COEX in Seoul on the 17th. It showed the future of cybersecurity in which AI detects and responds to threats in real time.
Mark Johnston, Google Cloud APAC's General Manager for Security, Networking and Collaboration, is giving a keynote speech at the '31st Information and Communications Network Information Security Conference (NetSec-KR 2025)' held at COEX in Seoul on the 17th. [Photo=Reporter Sojin Yoon]
General Manager Johnston mentioned the global IT outage caused by CrowdStrike last year, saying that not only malicious attacks but also 'mistakes' can be a major security threat. He said, "Last year, a single software update error brought down millions of computers," and "It was a mistake, not a malicious attack, but it raises questions about whether our digital infrastructure is really on a solid foundation." As
AI technology advances and cloud transitions accelerate, new security challenges are emerging. General Manager Stone warned that "AI is also used as a tool for hackers," and that it is evolving to automatically generate malware and evade security detection. Recently, new attack techniques such as 'Model Extraction' and 'Prompt Injection', which target AI models themselves, are also spreading.
General Manager Johnston explained, "Attackers are focusing on private and public cloud access points," and "Risks are increasing not only in open source-based AI models but also in the entire software supply chain and AI system supply chain. In a cloud environment, a single vulnerability can affect numerous systems simultaneously, so it is essential to strengthen the supply chain security management system."
Google's solution…
Google's strategy to respond to evolving security threats from zero trust to quantum computing is centered on the zero trust architecture. Zero trust is a security model based on the principle of "trusting nothing," and continuously verifies all access requests without creating boundaries between the inside and outside of the network.
General Manager Johnston said, "We introduced zero trust in 2009, allowing employees to work anywhere in the world without a virtual private network (VPN). This blocked the attack path itself," and "Since the distribution of physical security keys in 2017, there has not been a single case of account compromise due to phishing."
In the cloud environment, he emphasized ID verification between services and the "shared responsibility model." "We don't even trust services. We've built a system that assigns IDs to all services and verifies them. In a cloud environment, security is not a default, but a design issue," explained Johnston. "Google is responsible for the security of its infrastructure and services, but customers using the cloud should also take steps to secure their data and applications."
Google's future security strategy focuses on AI and quantum cryptography. First, Google is using AI as an accelerator for cybersecurity. "We perform 200 billion app checks and 5 billion malicious URL blocking on devices every day, which would be impossible without AI," said Johnston. In
order to prevent misuse of AI models and help introduce AI safely within organizations, the company is operating the 'Secure AI Framework (SAIF)'. It has also introduced the 'Agent-to-Agent (A2A)' protocol for secure connections between various agents.
It is also preemptively introducing quantum-resistant cryptography (PQC) in preparation for the era of quantum computing. General Manager Johnston added, "We have applied PQC to communications between data centers since 2023," and "We are strengthening cloud infrastructure security by applying it in addition to existing encryption."
This conference was hosted by the Korea Internet & Security Agency and organized by the Korea Information Protection Society, and was held under the theme of 'Strengthening Cybersecurity Collaboration through AI.' A total of 81 presentations will be held in 26 sessions until the 18th.
https://www.inews24.com/view/blogger/1835511
With the advancement of artificial intelligence (AI), cybersecurity threats are becoming more sophisticated, but at the same time, AI is evolving into a 'shield' that detects and responds to them in real time.
Evolving cybersecurity threats with the advancement of AI and cloud
Mark Johnston, Google Cloud APAC's General Manager for Security, Networking and Collaboration, demonstrated a future security system in which AI detects malware with a simple command in his keynote speech at the '31st Information and Communications Network Information Security Conference (NetSec-KR 2025)' held at COEX in Seoul on the 17th. It showed the future of cybersecurity in which AI detects and responds to threats in real time.
Mark Johnston, Google Cloud APAC's General Manager for Security, Networking and Collaboration, is giving a keynote speech at the '31st Information and Communications Network Information Security Conference (NetSec-KR 2025)' held at COEX in Seoul on the 17th. [Photo=Reporter Sojin Yoon]
General Manager Johnston mentioned the global IT outage caused by CrowdStrike last year, saying that not only malicious attacks but also 'mistakes' can be a major security threat. He said, "Last year, a single software update error brought down millions of computers," and "It was a mistake, not a malicious attack, but it raises questions about whether our digital infrastructure is really on a solid foundation." As
AI technology advances and cloud transitions accelerate, new security challenges are emerging. General Manager Stone warned that "AI is also used as a tool for hackers," and that it is evolving to automatically generate malware and evade security detection. Recently, new attack techniques such as 'Model Extraction' and 'Prompt Injection', which target AI models themselves, are also spreading.
General Manager Johnston explained, "Attackers are focusing on private and public cloud access points," and "Risks are increasing not only in open source-based AI models but also in the entire software supply chain and AI system supply chain. In a cloud environment, a single vulnerability can affect numerous systems simultaneously, so it is essential to strengthen the supply chain security management system."
Google's solution…
Google's strategy to respond to evolving security threats from zero trust to quantum computing is centered on the zero trust architecture. Zero trust is a security model based on the principle of "trusting nothing," and continuously verifies all access requests without creating boundaries between the inside and outside of the network.
General Manager Johnston said, "We introduced zero trust in 2009, allowing employees to work anywhere in the world without a virtual private network (VPN). This blocked the attack path itself," and "Since the distribution of physical security keys in 2017, there has not been a single case of account compromise due to phishing."
In the cloud environment, he emphasized ID verification between services and the "shared responsibility model." "We don't even trust services. We've built a system that assigns IDs to all services and verifies them. In a cloud environment, security is not a default, but a design issue," explained Johnston. "Google is responsible for the security of its infrastructure and services, but customers using the cloud should also take steps to secure their data and applications."
Google's future security strategy focuses on AI and quantum cryptography. First, Google is using AI as an accelerator for cybersecurity. "We perform 200 billion app checks and 5 billion malicious URL blocking on devices every day, which would be impossible without AI," said Johnston. In
order to prevent misuse of AI models and help introduce AI safely within organizations, the company is operating the 'Secure AI Framework (SAIF)'. It has also introduced the 'Agent-to-Agent (A2A)' protocol for secure connections between various agents.
It is also preemptively introducing quantum-resistant cryptography (PQC) in preparation for the era of quantum computing. General Manager Johnston added, "We have applied PQC to communications between data centers since 2023," and "We are strengthening cloud infrastructure security by applying it in addition to existing encryption."
This conference was hosted by the Korea Internet & Security Agency and organized by the Korea Information Protection Society, and was held under the theme of 'Strengthening Cybersecurity Collaboration through AI.' A total of 81 presentations will be held in 26 sessions until the 18th.
https://www.inews24.com/view/blogger/1835511
댓글
댓글 쓰기