Anthropic Building a ‘Pre-Crime’ System to Surveil Anti-AI Dissenters, Investigation Shows

Por Brad Reed — Common Dreams

The American Prospect on Wednesday published an investigation revealing that top artificial intelligence firm Anthropic is building a predictive surveillance system aimed at tracking activists who are opposed to advanced AI development.

The Prospect found that Anthropic last month issued a job posting seeking an intelligence specialist to “identify, assess, track, and investigate global threats including geopolitical instability, terrorism, crime, activism, nation-state targeting of the AI sector.”

This job will also employ a “pre-crime” approach that will involve “attempting to predict incidents before a crime occurs,” according to the Prospect. The concept of “pre-crime” was explored in Minority Report, a dystopian science fiction novel by Philip K. Dick.

A report last week in The San Francisco Standard revealed there is already at least one example of Anthropic taking this kind of predictive approach to policing.

Specifically, Anthropic called police last year and informed them that one of its users told the company’s Claude AI chatbot that he had purchased a rifle and had CEO Dario Amodei “in his sights.”

The AI lab told police that this person “was going to kill everyone at Anthropic,” but it didn’t provide law enforcement officials with transcripts of the chats, which would have let them make their own determination about the seriousness of the threat.

The person in question was not arrested or charged with a crime, and they told the Standard that they were “just fucking around” when they talked about buying a gun.

Eric Goldman, a law professor at Santa Clara University, told the Standard that he had some sympathy for Anthropic because if they don’t report potential crimes to police then “they face significant liability if, in fact, a crime does occur.”

“As they become more hated,” Goldman said of the AI companies, “there’s more pressure on them to over-disclose knowing that some of the people they identify for law enforcement shouldn’t be targeted at all.”

Anthropic isn’t merely using technology to monitor potential crimes, but also to keep tabs on US citizens lawfully exercising their First Amendment rights.

The Prospect tracked down a podcast interview with Keon Ellison, global security operations center manager at Anthropic, where he revealed how employees at the company use technology provided by risk detection firm Samdesk to monitor activists.

“Last year we had an executive travel into a major city when we received some intelligence through Samdesk about a planned protest,” Ellison said. “Samdesk gave us about 60 minutes of advanced notice that the protest organizers had moved the timeline… That extra hour was critical. Without it our executives would have departed their meetings, they would have ran right into the heart of the disruption.”

This story about taking great lengths to evade protesters drew a sarcastic reaction from tech journalist and author Brian Merchant.

“You know what you’re doing is good,” Merchant wrote in a social media post, “when you have to develop an in-house, real-time surveillance system to help your executives locate the ideal service exit to avoid contact with the public.”

Fonte: Common Dreams