Dark AI is no longer a looming threat — it’s here, and it’s powering some of the most sophisticated cyberattacks in the Asia-Pacific region, according to global cybersecurity firm Kaspersky.
Speaking at the 2025 APAC Cyber Security Weekend in Da Nang, Vietnam, Kaspersky experts warned that artificial intelligence, once hailed for productivity and innovation, is now being manipulated by both cybercriminals and nation-state actors to exploit digital systems at scale.
The term “Dark AI” refers to AI models — often large language models (LLMs) — that are used in ways that intentionally bypass safety, compliance, and ethical frameworks. These tools are now being deployed to create deepfakes, generate convincing phishing campaigns, write malicious code, and run cyber espionage missions without detection.
One disturbing trend: the rise of so-called Black Hat GPTs — AI tools like WormGPT, DarkBard, FraudGPT, and Xanthorox — designed or modified to automate fraud and facilitate cybercrime. These models are private or semi-private and thrive in underground circles, giving attackers unprecedented capabilities.
But what’s more concerning is the growing use of AI by nation-state advanced persistent threats (APTs). Kaspersky says attackers are embedding generative AI into covert operations that include crafting fake online personas, generating multilingual propaganda, and even responding to targets in real-time.
This revelation aligns with a 2024 report by OpenAI, which documented more than 20 state-linked operations attempting to misuse their models for influence campaigns and cyber exploits.
Kaspersky’s Sergey Lozhkin emphasized that AI itself isn’t inherently harmful — it’s the lack of control, oversight, and ethics that turns it into a cyberweapon. “We are entering an era where AI is the shield, and Dark AI is the sword,” he said.
In response to this new threat landscape, Kaspersky recommends that organizations adopt AI-powered security solutions, invest in real-time threat intelligence, control internal access to AI tools, and educate employees about the risks of shadow AI.
The future of cybersecurity now hinges on one core reality: only AI can keep up with AI.
Leave a Reply