Internal Security Shifts at Anthropic
Internal job postings and interviews with senior security officials indicate that Anthropic is constructing a broad monitoring system aimed at tracking activists who oppose the rapid development of artificial intelligence. Beyond simply monitoring protests near company properties, the lab is building a predictive surveillance framework to identify potential incidents before they occur. In certain instances, this involves reporting suspected individuals to law enforcement agencies prior to any actual criminal activity.
Anthropic did not respond to requests for comment regarding its surveillance infrastructure. This shift marks a departure from the firm's earlier attempts to position itself as a more cautious alternative to competitors like OpenAI. Internal documents show a trend toward treating domestic dissent as a national security issue. This aligns with recent corporate moves to secure defense contracts after a period of tension regarding the use of AI tools for mass surveillance.
The Role of External Monitoring Tools
Specific details of this surveillance architecture appeared in a 2025 podcast featuring Keon Ellison, Anthropic’s Global Security Operations Center Manager, and Zach Melvin, a Security Operations Manager. They discussed their collaboration with Samdesk, a company that provides real-time risk detection services. According to Ellison, Samdesk alerted Anthropic to a planned protest near an executive meeting, providing about 60 minutes of advanced notice that allowed security teams to reroute the executive through a service entrance.
This proactive approach aims to move beyond reactive measures. The goal, as described by an Anthropic security program manager, is to transform operations into a predictive and preventative threat management system. By using external data and internal monitoring, the company intends to maintain a defensible position for its high-value targets. This operational framework relies heavily on tools capable of gathering intelligence from social media and other open-source channels.
Investigations and Predictive Policing
Anthropic has begun making routine reports to police departments across the country. The company confirmed to The Wall Street Journal in July that it tracks concerning behavior over time through a person-of-interest process. This process is intended to catch escalation patterns early. In one notable case, Anthropic reported a man to the San Francisco Police Department after he told the AI model Claude that he had purchased an AR-15 and had CEO Dario Amodei in his sights. The individual later claimed he was joking, but the report had already triggered a police response.
An investigation by The San Francisco Standard found that Anthropic refused to share the actual messages with police, citing internal privacy policies. This creates a scenario where the company acts on internal speech analysis but withholds the primary evidence from law enforcement. The company is currently hiring an enterprise intelligence specialist with a salary range between $180,000 and $230,000 to conduct deep-dive research into activists, geopolitical instability, and emerging security trends.
Industry Context and Labor Relations
This expansion of surveillance coincides with efforts by industry groups like Americans for Responsible Innovation to have AI designated as critical infrastructure. Such a designation would legally place artificial intelligence on the same level as electricity and water. If successful, this change could grant tech firms deeper access to intelligence products from federal agencies. Critics argue this could lead to civic opposition against data centers being categorized as national security threats.
While the company invests heavily in predictive security, internal labor tensions persist. Security guards responsible for patrolling Anthropic’s campuses recently authorized a strike due to stalled wage negotiations. The company responded by instructing staff to work from home during the labor action. As CEO Dario Amodei acknowledges a crisis of trust between the public and the tech industry, the decision to monitor activists suggests that the company is prioritizing security operations over community relations. The long-term impact of this shift remains a point of contention for labor organizers and privacy advocates alike.