Google to ban chrome extensions that 'jailbreak' AI chatbots

Google is moving to crack down on Chrome extensions designed to bypass the safety restrictions built into artificial intelligence chatbots, a move cybersecurity experts say is long overdue.

The company is banning browser extensions that enable users to "jailbreak" AI chatbots, which attempt to override the safeguards built into platforms such as ChatGPT, Google Gemini, and Anthropic Claude and other generative AI services.

But beyond helping users bypass AI safety controls, many of these extensions have also raised serious privacy concerns, with researchers warning that some harvest large amounts of user data while disguising themselves as harmless productivity tools.

What are AI jailbreak extensions?

AI jailbreak extensions are browser add-ons that automatically inject prompts or hidden instructions into conversations with AI chatbots.

Their goal is to persuade the chatbot to ignore its built-in safety rules, allowing it to generate responses it would normally refuse, such as instructions related to cyberattacks, harmful content or other restricted topics.

Some extensions also promise to unlock "developer mode", remove content filters or generate unrestricted responses, making them attractive to users seeking fewer limitations.

While not all jailbreak tools are malicious, security experts warn that many require extensive access to users' browsing sessions and conversations with AI services.

Privacy concerns

Researchers say the biggest concern extends beyond bypassing AI safeguards.

Many Chrome extensions request permission to read and modify data across every website a user visits.

When installed, they may gain access to sensitive information, including chatbot conversations, browsing activity, login sessions and even personal or work-related information entered into websites.

Some extensions market themselves as AI productivity tools while quietly collecting user data for advertising, analytics or other commercial purposes.

Cybersecurity analysts have warned that users often approve these permissions without fully understanding how much information they are sharing.

Why Google is acting

Google's decision shows a growing concern over the rapid increase in AI-related browser extensions appearing on the Chrome Web Store.

As generative AI becomes more widely used in workplaces, schools and everyday life, browser extensions interacting with AI platforms have become increasingly popular.

However, security reviews have identified extensions that misrepresent their functionality, request excessive permissions or violate Chrome Web Store policies.

By removing extensions that facilitate AI jailbreaks or engage in deceptive data collection, Google aims to reduce opportunities for abuse while protecting users' privacy.

The crackdown also highlights a broader problem within the browser extension ecosystem.

Cybercriminals have increasingly used browser extensions to steal passwords, intercept browsing sessions, monitor online activity and harvest personal information.

Unlike traditional malware, malicious extensions often appear legitimate and are installed voluntarily by users seeking additional features or convenience.
Because extensions operate inside the browser, they can sometimes access far more information than users realise.

Impact on AI users

The changes are expected to improve security for millions of Chrome users, particularly those relying on AI tools for work, education and research.

Removing risky extensions could reduce the likelihood of sensitive conversations being exposed or misused. The move may also encourage developers to create AI tools that comply with privacy standards and clearly explain what user data they collect.

For organisations that have embraced generative AI, the policy could help reduce security risks associated with employees installing unverified browser extensions.

As AI tools become integrated into everyday digital life, companies are facing greater pressure to ensure that third-party applications do not undermine security safeguards or expose sensitive user information.

Cybersecurity experts recommend reviewing installed Chrome extensions regularly and removing any that are no longer needed or request unnecessary permissions.

Before installing an extension, users should check who developed it, read recent reviews and carefully examine the permissions it requests. If an extension promises to remove AI safety restrictions, unlock hidden chatbot capabilities or bypass platform safeguards, experts advise treating it with caution.