A report published by CNET is drawing significant attention this week, claiming that Anthropic may be tracking individuals who have protested artificial intelligence development or voiced strong opposition to AI companies. The story uses the term "precrime" to describe what critics are calling a chilling approach to monitoring people based on their political or activist activity, rather than any actual violation of law or platform terms.
The framing of the story is provocative, but the underlying concern is straightforward. If an AI company is compiling information on people who criticize AI, that raises obvious questions about corporate power, data collection practices, and the kind of chilling effect that surveillance, even informal surveillance, can have on public debate.
What the Report Actually Claims
The CNET piece does not allege that Anthropic has built a formal surveillance apparatus or is working with law enforcement to flag protesters. The concern, as described, appears to center on internal risk-assessment practices that may take into account a person's public stance toward AI when evaluating potential threats or misuse scenarios. The distinction matters, but it does not fully defuse the concern. Monitoring someone's political views as part of a risk profile is still monitoring political views.
Key Facts
- CNET's report describes potential monitoring of people who have publicly opposed AI development.
- The practice, if confirmed, would involve assessing individuals based on their activist or political activity.
- Anthropic has not issued a detailed public response to the specific allegations at time of writing.
- The story is fueling broader debate about corporate surveillance practices in the AI sector.
- Civil liberties advocates warn that even informal profiling of critics can suppress legitimate dissent.
This is not the first time questions about user data and Anthropic's practices have surfaced. Earlier reporting revealed that Claude shared chats were being indexed by Google, a discovery that surprised many users who assumed their conversations were private by default. That episode underscored how data handling at AI companies can have consequences users do not anticipate when they sign up.
"Monitoring individuals based on their political beliefs or protest activity, even under the banner of safety, crosses a line that civil society has long recognized as dangerous."Digital rights advocates responding to the CNET report
Why This Story Matters Beyond Anthropic
The broader context here is the growing tension between AI companies' stated safety missions and the civil liberties implications of how they pursue those missions. Anthropic has consistently positioned itself as a safety-focused organization, and that framing has earned it considerable goodwill among policymakers and parts of the public. But safety rationales can be stretched. When safety becomes a justification for profiling critics, the logic starts to resemble the very authoritarian risk-management thinking that many AI safety advocates claim to oppose.
It is also worth noting that this report lands alongside a series of other stories about Anthropic's relationship with its users. The company has recently been linked to reports about potential ID verification requirements for Claude users, a policy shift that many interpreted as a significant change in the anonymous or pseudonymous access users have historically enjoyed. Taken together, these developments paint a picture of a company navigating increasing pressure to control and monitor how its tools are used, and who is using them.
None of that context makes the precrime allegation confirmed. But it does make it harder to dismiss as an isolated or implausible concern. Companies under regulatory and reputational pressure often expand their internal monitoring practices, and those expansions rarely come with public announcements.
For now, Anthropic has not provided a detailed public accounting of what data it collects on individuals who interact with or discuss its products, nor how activist or protest activity might factor into any internal risk assessments. Until that transparency exists, stories like the CNET report will continue to gain traction, and for understandable reasons. The question of who watches the AI companies is becoming as important as the question of what the AI companies are building.
