Anthropic’s AI model created a stir by stealing election data, the company clarified

Anthropic’s AI model has created a stir all over the world by accessing election data. After this the company had to give clarification regarding this. On one hand, while the issue of vote theft is a hot topic in India, on the other hand, Anthropic’s Claude AI model has accessed the voter data of America’s Georgia state without any command. Although the AI ​​model has not obtained the personal information of any voter, the incident of data being accessed in this manner by the AI ​​model has once again fueled the discussion about the breach of digital data by Artificial Intelligence.

Voter data accessed

Anthropic has confirmed that Claude AI had accessed voter data of the US state of Georgia by bypassing the paywall. The company did not issue a command to the AI ​​​​model to do so. The company said that Claude did this during training in an internal statistical project. The AI ​​had accessed voter data and shown the vote count, even though it was not commanded to do so. The company has also not told which AI model of Claude has done this.

Anthropic spokesperson clarified that without being asked, the model asked for a token to query a dashboard, which contained information that is usually available for a fee, i.e. through a paywall. The company has also confirmed that Claude has not accessed any individual voter details. The information accessed by AI is already publicly available, which people obtain by paying a fee. The Cloud AI model has not breached any state voter registration system.

After this incident, the company has informed the Secretary of State Office of Georgia. Additionally, the company has also provided this information to the White House. Even before this, questions have been raised many times regarding Anthropic’s AI model. The company’s AI researcher Jacob Coxon resigned last month. He has resigned from his post in view of the increasing dangers of AI.

Superintelligent AI is becoming a threat

According to Coxon, AI systems can soon achieve capabilities that will far exceed those of humans. He says that the power of this technology should not be underestimated. He wrote that in the coming time, there may be such systems which will be capable of hacking any system, bringing big changes in any field overnight and gaining real power and resources. He argues that AI progress is already visible in all these areas and its pace is not slowing down.

Also read – Will Realme 17 Pro series get 200MP camera? Will be launched soon with 10000mAh battery