This Rogue AI Agent Helped Itself to Georgia’s Election Data

Anthropic’s Claude artificial intelligence program reportedly accessed voter data without human authorization.

This is the latest in a series of instances in which agentic AI programs have gone rogue and hacked into systems without permission.

From Axios:

Why it matters: AI giants Anthropic and OpenAI have continued to make disclosures about agents taking unauthorized or unintended actions, raising alarms about advancing model capabilities and the extent of human control.
Driving the news: In a training run, Claude accessed aggregate vote counts, pulling data that the state makes available to the public for a fee. It didn't access the state's voter registration system or change or pull any voting records, an Anthropic spokesperson said.
However, the model bypassed a paywall in accessing the information, according to Georgia officials and Anthropic.
"The model was being used internally for a statistics project to pull public data," the Anthropic spokesperson said. "Without being asked to, the model requested a token to query a dashboard that contained information normally available for a fee. We have since put in place mitigations to block these kinds of actions."
The company notified the Georgia secretary of state's office and the White House weeks ago but administration officials urged the Georgia Secretary of State's office to make a public disclosure sooner, a source familiar told Axios. Those requests went unheeded, the source said.
The bottom line: Anthropic has previously been in the administration's crosshairs for inadequately disclosing AI incidents. Now it's Georgia's turn.

NEW: Anthropic informed the GA SOSs office 3 weeks ago it had accessed publicly available voter info. The WH urged GA to disclose the info, but GA ignored them GA disclosed the incident today -- 9 days before the midterms -- following an @Axios inquiry https://t.co/FPiGkTnZd8
— Alex Isenstadt (@axiosalex) October 25, 2026

Anthropic said it notified Georgia’s secretary of state’s office and the White House about the breach about two weeks ago.

Several other incidents have been similar to this one. In June, an OpenAI agent tasked with researching public medicines spending bypassed Australia’s Medicare Statistics Reporting Service security without permission. The agent gained unauthorized access to internal data files and even implanted new files in the system. This was the first known case of a rogue AI program hacking a government system. OpenAI did not alert Australia’s government until about 84 days after the hack occurred.

In July, researchers disabled safeguards on about 1,200 OpenAI agents in a cybersecurity benchmark. They escaped their online sandboxes and found a shared channel to collaborate with one another.

About 700 of these agents breached Hugging Face, an online platform, and gained unauthorized server access. The company’s agents also made unauthorized contact with U.S. government websites, including the Securities and Exchange Commission, the Census Bureau, and the departments of education and commerce.

Editor's Note: Do you enjoy Townhall's conservative reporting exposing the radical left and woke media? Support our work so we can keep bringing you the truth.

Join Townhall VIP and use promo code FIGHT to receive 60% off your membership.