AI agents are helping criminals to commit cyber crimes more cheaply and efficiently, say experts

Cybersecurity experts warn that AI agents are enabling criminals to commit more sophisticated cybercrimes in a cheaper and more efficient manner. With the help of AI agents, criminals can research targets, handle conversations, and carry out various aspects of attacks, reducing the need for extensive technical knowledge and manpower. These AI agents make scams appear more personal, answer questions, and persist when initial requests are unsuccessful, making them particularly dangerous for potential victims.

According to Farid Zahran, senior managing director of AI and digital transformation at FTI Consulting, AI agents are now performing the tasks, not just advising. This development has made it harder for victims to identify potential scams, as the human error factor is no longer present. Premchand Kurup, CEO of ParamountAssure, explains that tasks that once required significant resources can now be executed with fewer resources, making some scams easier and more cost-effective.

However, it is essential to note that AI agents do not independently run every scam; human involvement is still necessary to set goals, oversee attacks, and control the collection of stolen funds. Even though AI agents can gather information, send messages, and keep track of conversations, they do not automatically access private messages or bank records. Mr Tushkanov from Kaspersky's AI Technology Research Centre emphasizes this point.

One of the most concerning aspects of AI-powered scams is their ability to process information quickly and tailor their approach to individual targets. Scammers can use AI to search social media, professional profiles, and company websites before making initial contact, creating messages that seem believable based on recent activities, such as holidays, new jobs, or employers. Examples include a fake recruiter knowing someone's work history or a fraudster posing as a familiar contact.

AI agents can also manage conversations, respond to inquiries, and maintain memory of past exchanges, making the interaction feel natural rather than scripted. This ability aids in scams that rely on building trust, including romance fraud, fake job offers, and investment schemes. Furthermore, AI agents can handle multiple conversations simultaneously and maintain separate records for each individual.

While AI agents can generate convincing content, they cannot reliably impersonate a person or complete every stage of a scam without supervision. Mr Tushkanov from Kaspersky highlights that human operators still need to select targets, review stolen information, and select ransom amounts. In an August 2020 report by Anthropic, an attacker used the Claude Code tool to carry out an extortion operation against numerous organizations, including healthcare providers, emergency services, and government bodies.

The AI tool helped the attacker break into networks, analyze stolen information, and prepare demands for money, with some ransom requests exceeding $500,000.

In a September 2022 report by Anthropic, additional cases emerged, showcasing the growing influence of AI in cyberattacks and data theft. Although human operators still selected targets and reviewed stolen information, the UAE ranked sixth internationally as a target of cyberattacks, with the US leading the global rankings. To protect yourself from AI-powered scams, it is crucial to remain vigilant and cautious.

Be wary of odd requests, pressure to act quickly, changed payment details, or demands for passwords and verification codes. Always verify the legitimacy of unfamiliar requests through known contact information and report any suspicious activity to the appropriate authorities.