Georgia’s top election official said on Oct. 26 that the artificial intelligence (AI) agent Claude developed by Anthropic was able to access publicly available data from a Georgia elections website but said that elections in the state are secure.
Georgia Secretary of State Brad Raffensperger said in a news release that the Anthropic AI agent accessed data from Georgia’s Data Hub, which it described as being free to the public.
“This isn’t news. Anthropic’s AI agent accessed information anyone can find publicly,” Raffensperger said. “Accessing public information is not a hack, and pretending otherwise is reckless and dangerous.”
The Republican official added that media reporting about the agent accessing the publicly available website is clickbait designed to drive engagement.
The Data Hub is meant to serve as a public transparency resource and is separate from the voter registration systems, his office said.
“Georgia is consistently recognized as having some of the most secure elections in the country, with voting systems not connected to the internet designed to prevent fraud, paper ballots, and risk-limiting audits,” Raffensperger said. “Twisting this into a security scare for clicks undermines confidence in our elections and is absolutely shameful.”
Anthropic, the developer of the flagship Claude AI, recently told media outlets that Claude accessed aggregate vote counts as part of a project and bypassed a paywall to obtain information.
“Claude didn’t access the state’s voter registration system or change or pull any voting records,” the company said in a statement to news outlets.
Anthropic earlier this month released a blog that said it had discovered its AI tool “working around a restriction to reach data that was gated by a token or a fee,” adding that it was able to access websites that used fees or tokens run by government agencies.
“We have chosen not to name the organizations involved in the examples below to avoid exposing vulnerabilities in their systems, and at their request,” the San Francisco-based firm said in the post. “For this reason, we also provide less detail about each case than we otherwise would.”
Anthropic “briefed the White House on these cases and notified each agency involved,” the blog said.
In Monday’s statement, Raffensperger’s office said Anthropic contacted his office to express concerns about its AI agent accessing the data on its Data Hub on Oct. 8, prompting an investigation.
“The White House contacted the Secretary of State’s Office about Anthropic’s AI agent,” his office said. “The Secretary of State’s Office shared the findings with the White House and made clear this was not the voter registration system though our public facing Data Hub.”
On Oct. 26, Anthropic told Raffensperger’s office that its agent did not access Georgia’s voter list or bypass any paywall, and that it accessed only information available to the public, according to the statement.
“Georgia has spent years strengthening election security while making election data more transparent and accessible to the public,” Raffensperger added.
Axios first reported the action by Claude on Oct. 25, according to the office.
Reuters contributed to this report.
We had a problem loading this article. Please enable javascript or use a different browser. If the issue persists, please visit our help center.