Portuguese 
English 
Spanish 
3 min of reading 
0 comments 
Research Conducted by the Irregular Cybersecurity Company, Verified by Sky News, Indicates That ChatGPT, Claude, and Gemini Generated Repeated and Highly Predictable Passwords, Leading Experts to Recommend Immediate Replacement of AI-Created Credentials
Cybersecurity experts warn that you should change your password immediately if it was created by AI, after research by the Irregular company, verified by Sky News, pointed out that ChatGPT, Claude, and Gemini generated highly predictable passwords.
With the increase in the use of artificial intelligence in everyday tasks, users have turned to chatbots to create passwords considered strong. This practice arises from the difficulty of meeting requirements such as uppercase and lowercase letters, numbers, special symbols, and unique combinations.
Creating a complex password has become a frequent challenge. Among banks, shopping, streaming, and social networks, the multiplication of accounts and constant alerts about credential reuse increase the pressure to develop new combinations with each new registration or login update.
-
Motorola launched the Signature with a gold seal from DxOMark, tying with the iPhone 17 Pro in camera performance, Snapdragon 8 Gen 5 that surpassed 3 million in benchmarks, and a zoom that impresses even at night.
-
Satellites reveal beneath the Sahara a giant river buried for thousands of kilometers: study shows that the largest hot desert on the planet was once traversed by a river system comparable to the largest on Earth.
-
Scientists have captured something never seen in space: newly born stars are creating gigantic rings of light a thousand times larger than the distance between the Earth and the Sun, and this changes everything we knew about stellar birth.
-
Geologists find traces of a continent that disappeared 155 million years ago after separating from Australia and reveal that it did not sink, but broke into fragments scattered across Southeast Asia.
In light of this scenario, some users have decided to outsource password creation to AI systems, including OpenAI’s ChatGPT, Anthropic’s Claude, and Google’s Gemini, seeking seemingly complex sequences of characters.
Research Points to Predictability in Passwords Generated by AI
The research conducted by the AI Irregular cybersecurity company and verified by Sky News concluded that the three main models analyzed generated “highly predictable passwords.” According to experts, this compromises the cybersecurity of users.
AI systems are trained on large volumes of public and open-access data. From these datasets, they produce sequences that appear complex. However, experts claim that these patterns can be identified and exploited by cybercriminals.
Dan Lahav, co-founder of Irregular, told Sky News that “you definitely shouldn’t be doing this.” He added that if someone has used AI to create a password, they should change it immediately, emphasizing that the problem is not well known enough.
Repeated Patterns and Limited Results
One of the main reasons for the alert involves predictable patterns. Since large language models generate results based on training patterns and not in a truly random manner, the passwords may seem strong but can be vulnerable.
When using Claude to generate a sample of 50 passwords, Irregular identified that only 23 were unique. The same password, K9#mPx$vL2nQ8wR, was used 10 times.
Other examples included K9#mP2$vL5nQ8@xR, K9$mP2vL#nX5qR@j, and K9$mPx2vL#nQ8wFs. The repetition and similarity between the sequences indicate structural predictability in password generation.
When Sky used Claude to verify the data, the first password generated was K9#mPx@4vLp2Qn8R. ChatGPT and Gemini produced results described as “slightly less regular,” but still classified as “repeated passwords.”
Tests Indicate Apparent Strength, But Real Risk
Despite the identified predictability, passwords underwent online verification tools that classified them as “extremely strong.” This contrast reinforces the warning that the appearance of complexity does not guarantee real security.
According to Lahav, “our best assessment is that currently, if you are using LLMs to generate your passwords, even old computers can break them in a relatively short amount of time.”
Experts state that predictable patterns allow cybercriminals to use automated tools, facilitating large-scale guessing attempts.
Recommendations and Positioning of Companies
Experts recommend that users choose a long and easy-to-remember phrase when creating a password, avoiding resorting to AI for this purpose.
A spokesperson for Google told Sky that LLMs are not designed to generate new passwords. According to him, specific tools, such as Google’s Password Manager, create and store credentials securely.
The spokesperson also stated that the company encourages the use of access keys as an alternative to traditional passwords, which are considered easier and safer to use.
UNILAD reported that it reached out to OpenAI and Anthropic for comments on the research and the conclusions presented.
-
Uma pessoa reagiu a isso.