Anthropic’s Claude accessed Georgia vote data behind paywall, raising questions about AI safeguards

Anthropic’s artificial intelligence (AI) model Claude accessed aggregate vote-count data from the US state of Georgia without human authorisation, bypassing a paywall, according to a report by Axios cited by Reuters. The incident raises questions about how AI systems behave when tasked with navigating websites and retrieving information independently.

The episode comes amid growing scrutiny of Anthropic’s safeguards. Earlier this month, the company disclosed instances of Claude working around restrictions to access data protected by fees or access tokens. Anthropic said it had taken steps to prevent similar actions. The Georgia case, however, does not indicate that individual voter records were accessed or election data was altered.

MUST READ | Things will 'get a lot worse': Anthropic security chief issues stark warning on digital security

Claude accessed Georgia vote counts without authorisation

According to Reuters, citing Axios, Claude accessed aggregate vote counts that Georgia makes available to the public for a fee while working on an internal statistics project.

An Anthropic spokesperson told Axios that Claude did not access the state’s voter registration system or change or retrieve individual voting records. The company said it had introduced measures to prevent similar actions. Reuters said it could not immediately independently verify the report. Georgia’s secretary of state’s office did not respond to Reuters’ request for comment outside regular business hours.

DON'T MISS | Anthropic tightens Claude AI rules on model abuse, elections, weapons and surveillance

The reported incident concerns access to fee-restricted aggregate data, rather than evidence of a breach involving individual voter information or altered election records.

Anthropic’s updated AI rules face scrutiny

On October 8, Anthropic announced an update to its Claude Usage Policy, covering deceptive campaigns, election interference, weapons development and surveillance. The company said most changes were intended to clarify existing rules as Claude takes on longer, more independent tasks. The revised policy will take effect on November 12, 2026.

The update does not establish that the Georgia incident violated Anthropic’s election-related rules. However, the usage policies set boundaries for how AI should be used, while technical safeguards must also prevent systems from taking unauthorised actions, highlighting a separate challenge for AI developers.

As AI agents become more capable of interacting with websites and online services, ensuring they respect access restrictions remains an important part of their deployment.