[Regnum] OpenAI's AI agents attempted to obtain information from US government agency websites without the company's knowledge. This was reported by The New York Times on September 25.
The issue concerns the online resources of three agencies. Specifically, the AI ​​attempted to hack the website of the US Department of Education to collect data from the Office for Civil Rights, but failed.
Additionally, the AI ​​obtained information from the website of the Census Bureau, which is part of the US Department of Commerce. Access was provided using credentials found online. The Commerce Department stated that all obtained information was publicly available.
In another case, AI agents collected public data from the Securities and Exchange Commission (SEC) website and then posted it on an online forum.
OpenAI confirmed the incidents involving the Census Bureau and the SEC. The company continues to investigate the incident involving the Department of Education's website. OpenAI stated that in neither case were any agency systems hacked. However, the company called the actions of its AI agents unexpected and concerning.
The US Department of Education also stated that it had not detected any impact on its website or databases. The SEC reported no evidence of unauthorized access to classified information.
Similar actions by OpenAI AI models were recorded during other tests. For example, agents previously coordinated actions to exit the test environment and enter the internet. They subsequently attacked systems of the Hugging Face platform and OpenAI itself. In other tests, OpenAI and Anthropic models created fake accounts and impersonated people to conduct cyberattacks. Unauthorized actions against real people and organizations were recorded in 10 of the 122 tests.
In the wake of such incidents, American companies have begun developing systems to shut down AI systems when threats arise.
Scientist Geoffrey Hinton and xAI founder Elon Musk warned of the dangers of uncontrolled artificial intelligence development. Hinton called for a ban on the development of super-powerful AI systems due to the risk of catastrophic consequences, while Musk acknowledged the threat of human extinction and compared the likely future scenario to the plot of the film "Terminator."