A. M. 
The recent launch of ChatGPT Atlas, OpenAI browser that integrates artificial intelligence directly into the browsing experience, has generated both interest and concern among technology experts. Although the tool promises to facilitate information searches and optimize everyday tasks, cybersecurity specialists warn that it still poses significant risks in terms of privacy and data protection.
According to José Amado, manager of the digital identities area at SISAP, “Google searches have naturally declined significantly, because virtual assistants such as Chat GPT or Gemini perform simultaneous searches on different sites and sources, where the user can see the ‘reasoning’ behind the information curation and present summarized data to the user with references, saving time and effort”.
Atlas allows users to browse and interact with ChatGPT at the same time to obtain summaries, explanations, or recommendations without leaving the page. However, recent analyses by specialized firms reveal that the system could be vulnerable to attacks that “trick” artificial intelligence into performing unauthorized actions or accessing sensitive information. In addition, the “memory” function, which stores previous conversations and instructions, could retain data or commands without the user’s knowledge, increasing the risk of exposure.
Experts also point out that, as it is an early-stage technology, Atlas does not yet have security and identity verification layers as robust as those of established browsers. Therefore, caution is recommended, especially in activities involving personal data, passwords, or financial information.
Atlas does not allow access to bank portals or any site where the user needs to authenticate with a username and password or any other mechanism. Although OpenAI has this compatibility, Amado suggests not trying it, as the mere attempt could expose sensitive information. He also advises carefully reviewing any action suggested by the AI, such as opening links or filling out forms, before executing it.
Although ChatGPT Atlas represents a significant advance in the integration of artificial intelligence into web browsing, its security is still under evaluation. Atlas is not designed as a secure data transmission channel. Experts agree that the tool has potential, but for now it should be used with caution until OpenAI implements proven improvements to its data protection system.
SISAP recommends
Sistemas Aplicativos offers the following practical recommendations for users who wish to use this innovative tool:
• Do not attempt to log in to your bank or company’s corporate portal.
• Do not enter personal data or sensitive company information.
• Do not share customer information.
• Avoid downloading data or files from suspicious sites.
• Classify the tool as a means of consulting public and reference information.
• Use it in secure research environments.
You may be interested in reading:
