Holiday Shopping 2026: When Bots, AI Crawlers, and AI Agents Converge


Every holiday season, retailers race to launch promotions and capture shopping demand, while consumers race to secure the best deals. Attackers on the other hand, race to exploit this surge in traffic for malicious activity, fraud, and abuse, targeting valuable accounts and sensitive transactions.

This year, there is a new participant in shopping activity on the internet: AI agents. This new class of traffic is growing exponentially - helping consumers research products, navigate purchasing decisions, and take actions on their behalf. Along with AI crawlers that continue to target retail sites for training models, indexing content, and retrieving information for real-time answers, retailers have to contend with a very different traffic mix compared to previous holiday shopping seasons. This means they can no longer only classify automated traffic as ‘good’ or ‘bad’ and manage them accordingly. They also need to determine who is visiting, understand their intent, and govern their actions.

The Last Holiday Shopping Season Delivered a Warning

Observations from our 2026 E-Commerce Bot Threat Report showed bad bots accounting for 43% of total traffic to e-commerce retailers protected by Radware, during the critical Cyber 5 period. Human shoppers represented 46% of total traffic, placing malicious bots within three percentage points of the entire human shopping audience.

Attacks also intensified across critical retail workflows. At one multinational retailer, account takeover activity increased more than fivefold, carding attacks increased by roughly fifteen times and fake account registrations grew sixfold compared to the previous season. Price scraping dominated the share of attacks with approximately five billion detected attacks. At another retailer, cart abandonment and denial-of-inventory activity rose 500% on Cyber Monday.

These attacks affect more than security operations. They directly impact business by draining inventory, compromising customer accounts, increasing infrastructure costs, exposing competitive pricing, and distorting conversion data.

The Threat Begins Before Black Friday

Last season showed that attacks can intensify in the days and weeks leading up to the biggest shopping events of the season.

At one retailer, account takeover attempts surged approximately 1,100% four days before Black Friday. Content scraping rose about 70% three days before the event. This corresponded with the period when customers were reactivating accounts, updating payment information, and e-commerce platforms were launching promotions, updating their catalogs etc.

The lesson for retailers is clear: their bot protection strategy and response policies should not wait for the holiday shopping traffic to peak but instead should be planned well in advance.

AI is Expanding the Automation Landscape

Nearly 70% of the bad bot traffic observed during the last holiday shopping season was classified as low in sophistication, compared with 44% a year earlier – indicating that the rapid adoption of Gen AI has collapsed the barrier to launch attacks and create attack tools. A large base of low-skill attackers are now able to leverage advanced AI tools and services to launch sophisticated, human-like bots.

Retailers need AI-powered bot management to fight against these advanced AI-enhanced threats. This includes AI-based behavioral detection across their applications, followed by effective mitigation of bot threats without affecting end-user experience. Radware Bot Manager combines preemptive protection, AI-based behavioral detection, and advanced mitigation techniques such as CAPTCHA-less crypto challenges to deliver real-time, AI-powered bot protection.

Not All AI Crawling Activity is Beneficial

During the 2025 holiday season, Radware observed that 54% of AI crawler activity originated from training crawlers, 27% from indexing crawlers, and 19% from retrieval crawlers.

Each of these AI crawling activities deliver very different business outcomes. Indexing and retrieval may help products appear in AI-generated recommendations. On the other hand, training crawlers often collect valuable catalog and pricing information to train LLMs, while consuming infrastructure and not offering value in return to the source.

Blocking every AI crawler could reduce a retailer’s visibility in AI-mediated discovery. Allowing all of them could expose proprietary content and consume valuable capacity that could better serve genuine customers. Hence retailers need granular control over AI crawling activity based on crawler identity, purpose, and business value.

AI Agents Require a New Trust Model

AI agents on the internet introduce an even more consequential challenge. A shopping agent may be automated, but it could represent a legitimate customer with genuine purchase intent. Blocking it simply because it is automated may mean rejecting the rapidly emerging agentic commerce revenue channel.

However, identifying an agent alone does not mean it can be trusted to execute any task on the application. An agent might deviate from its declared purpose, be manipulated, or its goal hijacked into carrying out unauthorized actions on sensitive workflows such as login, checkout etc.

Radware Agent Trust Management is designed to identify all AI agents in real-time, including standards-compliant agents and those that do not voluntarily identify themselves. It captures declared intent directly from the agent, evaluates whether behavior remains aligned with that intent and continuously assesses trust as new signals emerge.

This supports permission-based governance – a retailer could allow a medium or low-trust agent to browse products while restricting it from checking out and completing a transaction.

This holiday season, retailers can no longer rely on differentiating between humans and bots alone. Retailers that can determine the type of automated actor, evaluate its intent, and govern what it should be allowed to do will be the ones best positioned to capture the value of the agentic economy while containing its risks.

Dhanesh Ramachandran

Dhanesh Ramachandran

Dhanesh is a Product Marketing Manager at Radware, responsible for driving marketing efforts for Radware Bot Manager. He brings several years of experience and a deep understanding of market dynamics and customer needs in the cybersecurity industry. Dhanesh is skilled at translating complex cybersecurity concepts into clear, actionable insights for customers. He holds an MBA in Marketing from IIM Trichy.

Contact Radware Sales

Our experts will answer your questions, assess your needs, and help you understand which products are best for your business.

Already a Customer?

We’re ready to help, whether you need support, additional services, or answers to your questions about our products and solutions.

Locations
Get Answers Now from KnowledgeBase
Get Free Online Product Training
Engage with Radware Technical Support
Join the Radware Customer Program

Get Social

Connect with experts and join the conversation about Radware technologies.

Blog
Security Research Center
CyberPedia