Cheap AI-Driven Hacks Threaten Hundreds of Websites, Experts Warn

Security researchers have uncovered two distinct attack campaigns that exploit publicly available artificial‑intelligence models to compromise web properties at a fraction of the cost of traditional hacking operations.

In the first operation, adversaries leveraged open‑weight AI models to automate attacks against a catalog of 105 websites. By feeding the models with publicly accessible code and configuration data, the attackers were able to generate malicious payloads that bypassed common defenses, demonstrating how readily downloadable AI tools can be weaponized without the need for custom development.

A second, even more startling case involved the use of a public AI model to seize complete control of an entire domain for a reported price of just $12. The model was tasked with probing DNS settings, identifying vulnerable admin interfaces, and crafting the precise commands needed to overwrite records, effectively handing the attacker full authority over the targeted site.

Both incidents were highlighted in a recent Dark Reading report, which warned that the low barrier to entry for AI‑enhanced attacks could shift the balance of power toward opportunistic threat actors. The affordability and accessibility of these tools mean that groups without extensive technical resources can now launch sophisticated exploits that previously required specialized expertise.

Cyber‑defense teams are being urged to rethink traditional protective measures. Because the AI models generate novel attack vectors on the fly, signature‑based detection systems are less effective. Experts suggest augmenting defenses with behavior‑focused monitoring, threat‑intelligence feeds that track emerging AI tool usage, and deploying AI‑based analytics of their own to spot anomalous activity in real time.

Looking ahead, the security community anticipates a surge in similar low‑cost, AI‑driven campaigns. As more open‑source models become available and their capabilities expand, the line between hobbyist experimentation and organized cybercrime may blur further, prompting organizations to invest in adaptive, intelligence‑driven security frameworks before the threat landscape evolves beyond current safeguards.