A recent study conducted by cloud security startup Wiz and AI security lab Irregular reveals that the economic landscape of cybercrime is shifting significantly. AI-driven cyberattacks can now be executed for as low as $50, contrasting sharply with the traditional costs nearing $100,000 for human-led security assessments. This accessibility allows malicious actors to launch widespread attacks, raising concerns for organizations of all sizes.
During an interview, Gal Nagli, Wiz's head of threat exposure, and Omer Nevo, cofounder and CTO of Irregular, discussed their findings, which highlight the effectiveness of AI tools in performing complex security tasks. In controlled tests, AI agents managed to resolve 90% of modeled attack scenarios, surprising even experienced cybersecurity professionals with their rapid problem-solving abilities.
The study emphasizes a growing vulnerability as non-technical employees increasingly use user-friendly coding tools like Anthropic’s Claude Code and OpenAI’s Codex, often without adequate security knowledge. This trend expands the attack surface for cybercriminals, who can exploit systems with sensitive data exposed online. Although the research noted some performance decline in realistic conditions, it indicates a concerning trend where cyber threats are becoming both cheaper and quicker to implement.