AI mishap: ChatGPT accidentally exposes data of 3,000 Australian flood victims

In the latest misuse of AI, ChatGPT has leaked the data of up to 3,000 flood victims in New South Wales (NSW), Australia.

The government of NSW in an official statement on 6 October said the leak occurred when a former government contractor unadvisedly uploaded a spreadsheet with the data of flood victims on the chatbot. 

While no data has been made public according to the official statement, investigations are already on to determine the extent of damage.

Unauthorised action

The NSW Reconstruction Authority (RA) said the data breach affected some flood victims who applied for the Northern Rivers Resilient Homes Program (RHP).

This happened after a former contractor of the RA uploaded data containing personal information of the applicants to ChatGPT without prior permission in March.

Cyber Security NSW is already working with forensic analysts to monitor the internet and the dark web to see if any of the information is accessible online.

“We understand this news is concerning and we are deeply sorry for the distress it may cause for those who have engaged with the program. We will be contacting people this week with updates to let them know what has happened and whether they have been impacted or not.”

The data was a Microsoft Excel spreadsheet containing 10 columns and over 12,000 rows of information. Every row is being carefully reviewed to understand what information may have been compromised, the RA said.

Only up to 3,000 people have been confirmed to have been affected at the time of this report, but the forensic analysis is still ongoing to determine the full scope of the breach.

Data leaked include names and addresses, email addresses, phone numbers, and some personal and health information.

Bad side of AI

AI has been misused a lot recently, leading to catastrophic outcomes. Just recently, a couple ended a 15-year old marriage due to advice from ChatGPT.

The chatbot has also reportedly aided the suicide of a teeanager, forcing OpenAI to rethink its security policy for kids.