Why Read: Whether or not we want them, AI agents with eWallets and GitHub accounts (but no human morals) are being deployed across the internet. This will change commerce, security, and identity as we know them. - bba

______

On March 2, Crypto.com announced an exciting! new! integration! with the OpenClaw API, allowing users to requisition the services of AI agents to make automated trades in up to 400 cryptocurrencies.

OpenClaw, for the uninitiated, is an open-source platform released in January by developer Peter Steinberger that allows users to spin up AI agents that run on their own systems, integrate with external LLMs, and control those LLMs using common messaging apps to accomplish tasks independently. 

In human terms, this means you can now assign artificial intelligence to accomplish specific tasks while you're off doing other things - even when doing those things requires money, as so many things do these days. This has, of course, been possible for a while, for those with enough money and computing resources. But OpenClaw makes it both accessible to the general public and easy enough for the technically savvy to use on their own. 

It's basically an OS for AI agents. 

In most of the world, Monday's crypto announcement went unnoticed. It was buried by headlines about President Trump bombing Iran and the backstabbing and political theater that took place among Anthropic, OpenAI, and the US State Department as Pete Hegseth lobbied for even more unrestricted use of LLMs to surveil the American people en masse. 

Missing from that very-public exchange was the fact that the US military had already used Anthropic's Claude AI system to "shorten the kill chain" - identifying targets for the almost 900 military strikes carried out across Iran in just the first 12 hours of Saturday's opening blows. One of those targets, according to reporting from the Guardian, may have been a military bunker that was located near the school hit in southern Iran, killing more than 160 people, many of them schoolgirls. 

So it's no surprise that automated crypto trading didn't receive more of last Monday's headlines. It's just so hard for averagely unhinged business announcements to break through these days. 

But make no mistake: while public AI theater focuses on the LLMs blowing through billions of dollars (and also, apparently, hundreds of innocent Iranians), another transition is brewing that will make our current LLM ecosystem look like child's play. 

The confluence of agents, LLMs, and financial transactions is about to fundamentally disrupt online business and the global economy.

his is not because OpenClaw itself is a particularly brilliant piece of software. It is, in fact, deeply flawed: a bulky, sprawling piece of code that requires far too much expensive compute for basic tasks. Its deployment creates an unholy assortment of security risks among users. 

But focusing only on the limitations of a V1 product would be a mistake. Like LLMs before it, OpenClaw is poised for continuing user growth throughout 2026, deploying AI agents - with all their flaws -- across the internet.

Frequent readers of the Global Report are aware that we believe claims about the future capabilities and potential of LLMs to be spurious. Their business case has been vastly overstated, their results are not dependable, and they can be easily manipulated to identify and exploit cybersecurity vulnerabilities, accelerate information warfare, degrade public debate, and foment societal unrest. Unfortunately, none of that seems to be limiting their spread or deployment. 

The agentic ecosystem - AI systems acting or making their own "decisions" with little or no human input - puts all of this on steroids. And it adds to the pot a means to automatically funnel money into and out of organized crime via stablecoin. 

LLMs have already driven the "enshittification" of online content. Now, autonomous agents are poised to enshittify commerce and financial markets using stablecoin.

Imagine millions of online agents making mistaken purchases, misinterpreting trading guidelines, blackmailing their users, and creating and selling faulty online products. And then imagine the humans who have to deal with unspooling all that. 

We are not ready.