The rise of simply accessible AI applied sciences has dramatically lowered the bar for cybercriminals, enabling them to create and deploy malicious bots quicker and at higher scale than ever earlier than. Generative AI (GenAI) has made it easier to construct and refine bots, accelerating the evolution of automated threats. With AI’s assist, these bots have gotten extra subtle, extra quite a few, and much better at concealing their actions to evade detection.

Attackers at the moment are utilizing AI not simply to launch assaults, but in addition to investigate failed makes an attempt—studying and adapting in actual time. This fixed refinement makes AI-powered bots more and more elusive and harmful, posing a rising risk to organizations throughout each sector.

Automated site visitors now outpaces human customers

As automated exercise surges, safety groups face an pressing must rethink their defenses. They’re not combating solely human adversaries—bots at the moment are a dominant power on-line.

Based on the 2025 Imperva Dangerous Bot Report from Thales, titled “The Speedy Rise of Bots and The Unseen Threat for Enterprise,” automated site visitors surpassed human exercise for the primary time in a decade, accounting for 51% of all web site visitors in 2024. The explosive adoption of AI and huge language fashions (LLMs) has been the first driver behind this shift.

This rising wave of AI-driven automation has critical implications. As bots make up greater than half of on-line exercise, companies are going through mounting dangers from malicious automation that continues to escalate in each scale and complexity.

The Thales report additionally discovered that dangerous bot site visitors elevated for the sixth consecutive yr, representing 37% of whole net site visitors, up sharply from simply over 30% in 2023.

The rising sophistication of bot assaults

In 2024, “superior and average” bots made up 55% of all assaults, utilizing techniques that intently mimic human conduct and bypass conventional defenses. Detecting and stopping them has turn into considerably tougher as they mix into authentic site visitors.

On the identical time, easy, high-volume assaults surged—rising from 40% in 2023 to 45% of all bot exercise final yr. This development is basically fueled by the widespread availability of AI automation instruments, which permit even less-skilled attackers to execute bot-driven campaigns with minimal effort.

The Thales report additionally famous that 31% of all assaults it detected and blocked had been automated, aligning with the OWASP 21 Automated Threats classification, which targets net purposes at scale, exploits vulnerabilities, and evades safety measures throughout industries.

APIs: The brand new entrance line in opposition to dangerous bots

APIs at the moment are the spine of contemporary digital ecosystems, powering innovation, automation, and seamless integrations throughout providers. However that very same ubiquity has made them a favourite goal for attackers.

Thales’ knowledge exhibits that 44% of superior bot site visitors was geared toward APIs in 2024. Malicious actors exploit API endpoints to steal knowledge, commit fraud, or bypass controls.

“APIs are important to enterprise agility and innovation, however their inherent logic creates distinctive weaknesses that attackers are fast to take advantage of,” stated Chang. “As corporations increase their use of cloud and microservices architectures, it’s crucial to know that these identical strengths can introduce new dangers.”

Residential proxies: Disguising malicious site visitors

Cybercriminals more and more depend on residential proxies to disguise bot site visitors as authentic human exercise. By routing assaults via family IP addresses, they make it far harder for safety programs to inform pal from foe.

Thales discovered that 21% of all bot assaults leveraged residential proxies from ISPs, permitting risk actors to mix seamlessly with genuine customers and evade detection instruments.

AI-driven account takeover on the rise

Account Takeover (ATO) assaults have climbed sharply, up 40% yr over yr and 54% over the previous three years. This surge is linked to the rising use of AI and machine studying to automate credential stuffing and phishing—making assaults quicker, smarter, and tougher to detect.

The monetary providers sector bore the brunt of those ATO makes an attempt, accounting for 22% of all incidents, adopted by Telecoms and ISPs (18%) and Computing/IT (17%).

The double-edged sword of generative AI

Generative AI platforms resembling ChatGPT, ByteSpider Bot, ClaudeBot, Google Gemini, Perplexity AI, Cohere AI, and Apple Bot have reworked how folks study, work, and create. However they’ve additionally launched a brand new wave of cyber danger.

Whereas these instruments improve productiveness, risk actors are exploiting them as highly effective new assault vectors. Notably, ByteSpider Bot accounted for a staggering 54% of GenAI-enabled assaults, highlighting how AI’s democratization is reshaping each innovation and cybercrime alike.

Go to us to obtain your copy of the 2025 Dangerous Bot Report.