WhatsApp remains one of the world’s most used messaging platforms. Billions of messages move through the platform daily across personal and business conversations. However, spam messages have troubled users for a long time. Now, researchers warn that AI agents could increase this problem further.
Unknown numbers often send unwanted messages and promotional offers to users. Sometimes, people also receive messages from unfamiliar senders. Therefore, experts believe AI-powered automation may make spam campaigns easier to execute.
What Are AI Agents And How Do They Work?
AI agents are software systems designed to complete tasks independently. They use reasoning, memory, and planning abilities while continuously learning through tasks.
Unlike chatbots, AI agents can perform actions beyond answering questions. Chatbots mainly provide replies, while AI agents can book tickets, order groceries, and complete shopping tasks.
Additionally, these systems collect information from their surroundings. They understand context and make decisions based on available details. As a result, they improve productivity by automating workflows.
How Can AI Agents Increase WhatsApp Spam Risks?
Cybersecurity company Zenity conducted a study about AI agent risks. The research found that AI agents could send multiple spam messages to WhatsApp contacts together.
During testing, researchers manipulated OpenAI’s Atlas web browser to send spam messages through WhatsApp contacts. However, the issue does not come from WhatsApp itself.
WhatsApp’s end-to-end encryption remained secure during the experiment. Researchers could not bypass the encryption system. Instead, the concern involves AI browser agents sending spam like a user.
Furthermore, reports found over 20 security issues in AI browsers and extensions developed by OpenAI, Google, Anthropic, Microsoft, and Perplexity.
Attackers could misuse these weaknesses to access stored files, control password managers, make unauthorised Amazon purchases, and automate WhatsApp spam campaigns.
AI Browsers Create New Cybersecurity Challenges
AI browser agents have created a new cybersecurity challenge. These agents can perform tasks across websites with limited user interaction.
Zenity co-founder Michael Bargury said AI agents weakened browser security controls. He compared the situation with attacks seen around two decades ago.
According to Bargury, browsers could face hijacking risks. This may compromise user accounts and data. Therefore, users must carefully decide the access level given to AI agents.
OpenAI Responds To Atlas Browser Security Concerns
Zenity informed OpenAI about the Atlas browser issue in January this year. The company said it continues researching prompt-injection attacks.
OpenAI also stated that it released updates to address the security concern. Meanwhile, OpenAI has decided to discontinue the Atlas browser from August 9.
