ChatGPT Atlas and cyberattacks: how the new OpenAI tool could be the perfect weapon for phishing

The release of ChatGPT Atlas, OpenAI's new AI-powered browser, has aroused both enthusiasm and concern. A recent report warns that its high level of automation could make it up to 90% more vulnerable to phishing attacks, making it a possible target for the most sophisticated cyberattacks. In this blog we analyze what risks it poses, what lessons it leaves for cybersecurity and what we recommend in the face of this new generation of digital threats.

‍

ChatGPT Atlas: the innovation that alerts experts

The ChatGPT Atlas browser, developed by OpenAI, promised to revolutionize the way we navigate: automation, intelligent summaries and commands executed by AI. However, a recent study warns that it could be up to a 90% more vulnerable to Attacks of Phishing than traditional browsers.

‍

The problem lies in their own strength: autonomy. Atlas not only displays web pages, but it interprets and executes instructions, which opens the door to prompt injection or CSRF attacks, capable of manipulating the actions of the assistant. Even worse, those malicious commands can persist in ChatGPT's memory, even if the user changes session or device.

‍

Why should we be worried?

Because ChatGPT Atlas combines three key risk factors:

• Unlimited automation: the smarter, the more attack surface.
• Data persistence: Malicious commands can “survive” logging out.
• Overtrust: Many users assume that “if it's OpenAI, it's safe”.

This balance between innovation and exposure reminds us that even the most advanced technologies can become vectors of cyberattacks if they are not designed with security from the start.

‍

A New Scenario for Business Cybersecurity

The ChatGPT Atlas situation reflects a broader challenge: the boundary between productivity and digital risk is blurring. More and more companies are integrating AI solutions into their processes — from customer service assistants to automated analysis tools — without an in-depth assessment of their vulnerabilities.

This trend opens up a new attack surface: language models can be manipulated through malicious prompts or falsified training data, allowing attackers to influence automated responses or decisions. In corporate scenarios, this could lead to leaks of confidential information, improper access or manipulation of internal communications.

Therefore, AI security must be treated as an essential pillar within cyber defense strategies, not as a technical complement. Organizations that integrate these technologies must do so with continuous oversight and clear control policies over data flow and automation permissions.

‍

Key Lesson: Resilience Matters

The ChatGPT Atlas case demonstrates that cybersecurity resilience—anticipating, resisting, and recovering—must be a priority before adopting any AI-based technology.

Companies and users should review permissions, limit automation, and educate about the safe use of intelligent tools. Innovation without security isn't progress: it's risk.

‍

Tips for making responsible use of AI

At Apolo Cybersecurity, we believe that the evolution of tools such as ChatGPT Atlas represents a new era in the interaction between humans and artificial intelligence, but also a critical point for digital defense.

Our position is clear:

• The secure adoption of AI must be backed by ongoing audits and data control policies.
• Cyberhygiene training is essential: understanding how threats act is the best way to prevent them.
• Every intelligent tool must be designed under the principle of “security by default and privacy by design”.

The future of cybersecurity is not about avoiding AI, but about integrating it in a responsible and resilient way. At Apolo, we support organizations that want to innovate without compromising their security.

‍

Is your company prepared for the risks of AI?

Artificial intelligence drives innovation, but also new risks of cyberattacks and phishing. At Apolo Cybersecurity, we help you assess vulnerabilities, protect your data and strengthen your digital resilience against emerging threats such as those revealed in the ChatGPT Atlas case.

‍