Concealed within his fortress, the lord of Mordor sees all. His gaze pierces cloud, shadow, earth, and flesh. You know of what I speak, Gandalf: a great Eye, lidless, wreathed in flame.
-From Lord of the Rings, on the Eye of Sauron
Anthropic’s CEO, Amodei, released a statement opposing the Pentagon’s terms that would enable mass domestic surveillance and the use of autonomous weapons. Amodei recognizes the extreme peril in combining these technologies with current laws that are woefully inadequate to protect us. If you haven’t read Amodei’s statement, do so.
This story is making headlines; “Anthropic Just Showed What Doing the Right Thing Looks Like.” But what’s utterly absent from these headlines is that OpenAI, Google, and xAI have willingly allowed their AI to be used without safeguards by the Hegseth Pentagon putting all of us in peril.
OpenAI, Google, and xAI have willingly allowed their AI to be used without safeguards by the Hegseth Pentagon putting all of us in peril.
The media’s handling of the degree of danger this issue presents is lacking. AI Surveillance and Autonomous Warfare is one of the most immediate, catastrophic risks humanity faces, exceeding that of the testing of the atom bomb.
My assessment of the threat level imposed by this issue is based on empirical knowledge working with AI, intuition derived from closely following what unfolded in Israel-Gaza where these intelligence and automated weapons were tested, and my observations of recent media releases. I’ll break down the possibilities to provide those without much exposure to AI’s capabilities with a realistic picture of the peril that is on the horizon unless we act to address it.
My goal is to raise awareness and initiate a public conversation rooted in compassion and shared humanity about the very real threat this poses.
The Patriot Act enabled federal agencies to conduct broad surveillance with reduced judicial oversight, significantly impacting privacy. Key provisions still active or modified include using National Security Letters (NSLs) to obtain records without a warrant, “sneak and peek” secret searches, and accessing diverse records (business, library, financial).. However, with advancements in AI, these can be parsed profoundly quickly in a way that allows the creation of far-reaching profiles of individuals’ entire digital histories in moments.
Let this sink in.
OpenAI, Google, and xAI have APIs that have access to thousands of companies’ data – nearly every software tool you’ve used is now connected to an API at one of these companies. This means that an AI has touched and indexed that data: every email, document, search, website you’ve visited, text message sent, contact you’ve entered, social media post you’ve made. Every file on a Microsoft device if you’ve enabled with Copilot, every file on an Android with Gemini enabled. A digital trail spanning the last two decades at least can be assembled in an instant, mapping out every correspondence within your social network and every intimate thought you’ve documented in writing that you assumed would be kept private.
At the beginning of 2026, we saw a surge in subpoenas aimed at tech companies administered by the DOJ for people’s data. We’ve seen requests for voter registration data. The IRS provided 42k+ pieces of data illegally to intelligence agencies. This allows those in power to combine voter data, social media posts, SMS/text messages or any unencrypted chats, geolocations, and images of people into enormously detailed dossiers, with AI enabled search capabilities to surface all connected pieces of information in seconds. This data can easily be used to build networks of association along ideological lines. This can be deployed silently, and “legally” to systematically oppress, deny freedoms, and track anyone.
When you combine this with the capability to instruct machines to cause harm you have the “legal” ability to harm anyone perceived as an ideological threat with no accountability linked to any human because a machine pulls the trigger. We’ve seen the precedent of how far this administration is willing to go in abusing technologies and legal safeguards already.
Most people are only minimally aware of the rapidity with which a panopticon of unprecedented reach and precision could be created.
Anthropic is the ONLY company standing up to this threat while OpenAI, Microsoft, Google, and xAI have quietly offered their services for the building of weapons that no human should ever have the capacity to wield.
This should cause great concern. We need to collectively raise visibility on this issue, and push Congress to reign in this massive overreach by elevating the visibility of these risks.
We’ve seen time and time again that commitments to safety, or contractual provisions highlighted in placating PR campaigns by public & private companies are hollow and can be changed silently within days following a press release. This provides no guarantee to the American people that protections are in place. The only way to ensure these protections is through a system of checks and balances provided by an oversight panel that builds law proactively to prevent problematic use cases as these projects are underway. Current lax privacy laws with the Patriot Act enabled the scenario described above unless Congress legislates otherwise.
We need Congressmembers to create an oversight panel that reviews any and all contractual language relating to uses of AI with domestic data and proactively creates new legislation that protects people from problematic abuses of power that are currently enabled by US laws including the Patriot Act and now accelerated by AI. These laws need to have stiff, deterrent penalties, on the order of decades in prison, for anyone giving instructions to humans or AI to assemble AI tools capable of compiling dossiers from the private data of any person without judicial review and notification of the target or for anyone using a tool of this nature, in order to prevent abuse and protect people.ect people.
We are at a precipice and the public is mostly unaware of just how pivotal this moment in time is. If anyone has that kind of visibility into other’s lives without their knowing, it would be terrifying. Now consider the people currently in Washington having that visibility into every digitally connected person’s life — it’s the equivalent of the Eye of Sauron, through which our leaders can view and locate any suggestion of ideological dissension, be it a simple “like” on a post that opposes illegal policies to something more complex — and thereby, extinguish that voice through whatever means they deem appropriate.
We’ve seen the distorted and deadly syntax for which this administration defines as reasonable force used against perceived “terrorist” threats.
This level of omniscience, in the hands of anyone let alone our current leaders, should never be allowed. This Pandora’s Box must not be opened.
Please take a few moments to inform your Congressional representatives of this issue and take action to protect us.
