Cybersecurity researchers at SentinelOne’s SentinelLABS have documented AkiraBot, an AI-powered spam automation framework that targets website contact forms, comment sections, and chat widgets. Active since September 2024 under the name Shopbot, the tool has targeted over 420,000 websites and successfully delivered spam content across approximately 80,000 of them. It uses OpenAI’s gpt-4o-mini model to generate custom, context-aware spam messages for each target — effectively bypassing content filters that rely on detecting repeated or template-based text.
Technical Architecture and AI Integration
Built using Python, AkiraBot represents a significant evolution in spam automation technology. The malware specifically targets contact forms, comment sections, and chat widgets commonly found on small and medium-sized business websites. What sets AkiraBot apart is its integration with OpenAI’s language models to generate unique, context-aware spam content that effectively bypasses traditional detection methods.
Operational Sophistication and Target Selection
Initially deployed in September 2024 as Shopbot, the malware has expanded its scope beyond its original Shopify-focused attacks. The threat actors have broadened their targeting to include websites built on major platforms such as GoDaddy, Wix, and Squarespace, with particular attention to sites utilizing Reamaze widgets. The malware leverages OpenAI’s gpt-4o-mini model, configured to function as a marketing message generator.
Advanced Security Evasion Capabilities
AkiraBot demonstrates remarkable proficiency in circumventing security measures. The malware successfully bypasses multiple CAPTCHA implementations, including hCAPTCHA, reCAPTCHA, and Cloudflare Turnstile. To maintain stealth, it employs the SmartProxy network infrastructure and implements sophisticated user behavior simulation to avoid detection.
Performance Tracking and Campaign Analytics
The malware maintains detailed operational metrics through a submissions.csv file, documenting all spam deployment attempts. Performance statistics regarding security bypass success rates are regularly published to a dedicated Telegram channel, enabling operators to optimize their attack strategies in real-time.
Small business websites with public contact forms as primary targets
Small and medium-sized business websites are the primary targets — particularly those using Shopify, GoDaddy, Wix, Squarespace, or similar platforms with publicly accessible contact forms or live chat widgets (such as Reamaze). The spam being deployed promotes dubious SEO services, but the same delivery infrastructure can be repurposed to distribute phishing links or malware. OpenAI disabled the API key and associated assets after being notified by SentinelOne, but the framework can be reconfigured to use other AI providers.
Protecting your website from AkiraBot spam
- Implement advanced CAPTCHA — hCAPTCHA, reCAPTCHA v3, or Cloudflare Turnstile; AkiraBot bypassed earlier CAPTCHA versions, so use the latest configurations and consider behavioral challenge layers.
- Rate-limit form submissions by IP and user agent, and add honeypot fields that bots fill but humans leave empty.
- Use Cloudflare or a WAF to filter known malicious IP ranges and block automated form traffic.
- Monitor for SmartProxy and datacenter IPs in form submission logs — AkiraBot routes traffic through residential and datacenter proxies to evade geo-blocking.
- Review your AI provider’s usage policies: AkiraBot demonstrates that any LLM API accessible without strong safeguards can be weaponized for spam generation at scale.
