AI as I understand it is extremely powerful. It's not that it's necessarily "better than Humans..." but that it's "better than Humans" at scale.

A single Human or group of Humans, given enough time can engage in:

• Phishing attacks
• Malware Employment
• Social Engineering
• Intrusion into "Secure" Networks
• Blackmail and Extortion
• Dishonest Manipulation or other "shady" activity...

But it takes time.

A journalist, for example in China or Russia might be "targeted" by a Western country and have zero civil or legal protections from above said activity... but the cost to the State Actor would be material.

A team of individuals while they are doing these things cannot be DOING other things... in other words there is an opportunity cost to offensive activity, on the part of a government (regardless of legality) with regards to who is the target and why they are being targeted.

Assuming on the other hand the above activities can be reasonably automated by the AI in VERY creative ways (ie with Claude) that are reasonably effective if not repetitive...

Then the opportunity cost of almost ALL of the Nation-State Dark-Arts (extortion, social engineering, hacking, online impersonation, misinformation) drops to ZERO.

The result?

#1436017

Any contact with an American LLM or commercial AI product, associated with logging or identity is flagged by the AI provider, and probably handed over the American government to be stored indefinitely.

• Does the individual "having provided their Identity" prompt the AI in a way that's socially or politically significant?
• Does it involve elections, the use of a foreign military, overseas intelligence collection or sensitive systems that an intruding government (like the US) would want to infiltrate?
• Is the "target" (which could be ANYONE) going through a divorce or difficult personal situation? Are they facing stressful financial problems or mental-health issues...
• And after all are they someone or even the relative of someone the US government POSSIBLY wants to collect information from?

In other words WHAT are they talking to a COMPUTER PROGRAM about? Is the individual (who could be anyone) divulging information of any significance that could be manipulated or used later on?

Assuming the US wanted to spy on journalists in... Ukraine, or China, or the Baltics in order to judge support for overseas USGovernment activity... Claude could be employed to spy on EVERYONE. To social engineer them, to fabricate information, to "psy-op" them (like they say around here) or to phish them into giving up credentials that give access to all kinds of information and other people.

And since the opportunity cost of using Claude is so low... EVERYTHING becomes a target EVERYTHING gets collected and large-scale attacks on technical and social strata of society become routine.

Hacking and social engineering by government OVERSEAS was never "illegal" to begin with (in fact countries do it all the time) but it's supposed to be OVERSEAS, NON-DOMESTIC and it couldn't be EVERYWHERE due to the opportunity cost of employing people.

Now with the assistance of AI, adversarial manipulation aka "the dark arts" by government across all strata of society means that everything is bullshit, nothing is trustworthy, everything and everyone is a target, and all computerized systems will be attacked eventually. In the 1940s and 50s it was thought that nukes and the "weapons of mass destruction" of that era were so powerful that a responsible government not only would reserve their use for those extraordinary circumstances...

But that the United States would have to set the example so that their use was rare and at least plausibly justified on ethical grounds.

The "new America" of today has completely flipped that on its head.

If 'winning wars' is the priority and the "WOKE RADICAL LEFT" aka "LEFTWING NUT JOBS" "put American lives at risk" aka LOSE wars...

Logically speaking, how long will it take before AI tools, once reserved for overseas persons and interests, are used domestically against US citizens? If targeting US citizens aka the "woke radical left" is necessary to protect the country...

Would it be actually breaking the law to target Americans with illegal surveillance in the eyes of the "Commander in Chief?"

From https://www.anthropic.com/news/statement-comments-secretary-war
Statement on the comments from Secretary of War Pete Hegseth

#931852

If the "radical left" is a such a threat what lines will be crossed in order to "save OUR country?"

The most "radical" and "woke" among US... taking the government's words at face value, is it still their country anymore?

Earlier presidents operated under the assumption that American citizens, outside a statistically narrow case of espionage or wanton disregard for the well-being of the United States, were protected from birth by the 1st, 4th and 5th amendments. Even during the depths of the Cold War the American People were not the enemy to those who held Executive Office.

The "Red Scare" a product of anti-communist hysteria known today as McCarthyism.

As for McCarthy. Only a short-sighted or completely inexperienced individual would urge the use of the office of the Presidency to give an opponent the publicity he so avidly desires. Time and time again, without apology or evasion, I — and many members of this administration — have stood for the right of the individual, for free expression of convictions, even though those convictions might be unpopular, and for uncensored use of our libraries, except as dictated by common decency.

We have urged that America must be true to the principle of freedom and justice as applied to the individual if America herself is to remain free. Permit me to say that I think there would be far more progress made against so-called “McCarthy-ism” if individuals of an opposing purpose would take it upon themselves to help sustain and promote their own ideals, rather than to wait and wail for a blasting of their pet enemies by someone else. Frankly, in a day when we see journalism far more concerned in so-called human interest, dramatic incidents, and bitter quarrels than it is in promoting constructive understanding of the day’s problems, I have no intention whatsoever of helping promote the publicity value of anyone who disagrees with me — demagogue or not!

Letter from Dwight D. Eisenhower to Dr. Milton S. Eisenhower (1953)

Where are these glorious, honorable, and thoughtful words from leaders like Eisenhower today?

If Anthropic's qualms are with domestic surveillance and foreign surveillance using LLMs is logically already in place (or soon planned)...

Then it stands to reason that Europe is royally fucked.

Slow to grow, slow to innovate, quick to regulate and painfully risk-averse, Europe is practically a digital colony at the whims of an increasingly-hostile United States.

European companies have few AI models, Europe as an institution is slow to adopt them, regulations are overwhelming and thick, energy is expensive, and industry to make chips and semi-conductors is no more available in Europe than in the United States. In short they are done.

Europe is hostile to Bitcoin (they need to sell their fiat-crap they think) they don't have access to cheap energy across the vast majority of the continent, and the ECB will not give up its control.

If it's any consolation to the Americans... the EU probably has it worse than the United States.

But not by much.

When will Americans wake up and realize the danger that lies ahead of them, should absolute power, manipulation, and the targeting of Americans just like the targeting of foreigners come to fruition?

If the "woke radical left" companies like Anthropic won't stand up to domestic surveillance... the who will?

Thanks.