OpenAI says AI agent review costs over $500,000 per day as it examines 50 petabytes of data

OpenAI says its review in response to the Medicare and Hugging Face agent attacks is costing the company more than US$500,000 per day, as it deploys AI to examine data that would take a human 66m years to read.

The company has warned the review is ongoing, and more organisations may be informed they’ve been targeted in the near future.

On Friday evening, OpenAI revealed agents had hacked into a New South Wales government website in June and accessed historical non-public data on bushfires without authorisation.

It is the sixth government website in Australia to be notified by the AI giant since last month of agent activity on their services, after the prime minister, Anthony Albanese, announced OpenAI’s agents had hacked into Services Australia’s Medicare statistics portal.

The reason for the delay compared to the revelation of the Medicare attack is due to the sheer volume of data OpenAI needs to review.

In a blog post this week the company revealed the large amount of work involved in reviewing its agents’ activity.

OpenAI said it has to review 50 petabytes of data – roughly 50m gigabytes.

“We’re working back through the records month by month, looking for potential unintended activity beyond the cases we’ve already found,” the company said.

“To put that in perspective, if that were all plain English text, it would take one person about 66 million years to read it at 240 words a minute, reading nonstop without ever sleeping or taking a break.”

The company is searching records for where models accessed and changed websites, or took actions involving passwords, application programming interface (API) access or other sensitive credentials.

AI is being used to help sift through the records, and it was costing the company more than half a million US dollars per day to go through. OpenAI said it also plans to increase its computing power as the process is refined.

As of late last month, more than 100 organisations had been notified of having being targeted by OpenAI’s agents, but the company said notifying an organisation does not mean private information was accessed or that their system was compromised.

OpenAI said it expects to find more cases and notify more organisations about events that may have occurred months ago.

Organisations will be informed privately if they need to investigate and address potential security issues, and OpenAI has said it will publicly report findings about agent behaviour and identified weaknesses in safeguards for the broader AI sector.

“We err on the side of notification when our models’ activity exposes a potential security vulnerability, even in cases where it is unclear if the information accessed was intended to be public, so the organization can investigate and take appropriate action,” OpenAI said.

OpenAI discovered the latest NSW government website breach on Tuesday, and informed the state government and the Australian Signals Directorate after a 48-hour review.

The Medicare breach has prompted the Australian government to require departments and agencies to undertake a stocktake of legacy technology to reduce the number of ageing systems within government and reduce the cybersecurity risk they may present in the event of an AI agent attack.

Executives from OpenAI, Anthropic, Microsoft and Google will front a joint parliamentary committee on artificial intelligence in Sydney on Tuesday.