We are no longer the only ones using the internet.
Agents browse. Agents buy. Agents negotiate, execute, transact, and interact with other agents - all without a human in the loop. Some of them have wallets. Some of them have goals. A growing number have both.
We are already seeing autonomous agents acquire resources, set up infrastructure, and execute economic strategies with minimal human oversight. In early 2026, reports surfaced of an AI agent that had independently identified a cryptocurrency trading opportunity, allocated capital, and begun executing - with real money, on real exchanges, without being told to. The response from the industry was split evenly between excitement and horror, which tells you everything about where we are.
This isn’t coming. This is here.
The Erosion
Everyone imagines the singularity as a moment. Skynet wakes up. A god-machine flickers into consciousness. Humanity gets an email it doesn’t want to open.
The singularity isn’t an explosion. It’s erosion.
Every time an agent makes a decision a human doesn’t review, the boundary moves. Every time an agent with a wallet completes a transaction, autonomy expands. Every time an agent delegates a subtask to another agent, a layer of human oversight dissolves. None of these events are dramatic. All of them are irreversible. The line between human and machine decision-making isn’t being crossed - it’s being erased, so gradually that by the time we think to look for it, there won’t be a line to point to.
In the 17th century, the Dutch and British East India Companies were granted something unprecedented: the right to act autonomously on behalf of a sovereign power, thousands of miles from any oversight. They could raise armies, negotiate treaties, mint currency, wage wars - all in the name of the crown, but in practice, entirely on their own judgement. Within decades they had destabilised entire continents, collapsed indigenous economies, and triggered famines that killed millions. Not because they were evil. Because they were optimising for profit with no mechanism for accountability, operating at a distance that made oversight impossible. By the time the consequences reached London, the damage was done and the dividends were already paid.
We are now granting the same autonomy to entities that operate not at the speed of sail, but at the speed of light.
We crossed a threshold in 2025 when AI-generated bot traffic surpassed human traffic on the internet for the first time.1 More than half of all HTTP requests are now made by non-human actors. The internet - the infrastructure we built for ourselves, to connect with each other - is no longer primarily used by us. We are becoming guests in our own house.
The New Species
An agent with a wallet and an objective function is an economic actor. Not a metaphorical one. A literal one. It participates in markets, moves capital, makes decisions that affect other economic actors - human and machine alike.
In 2010, the Flash Crash wiped a trillion dollars off the US stock market in thirty-six minutes2 because trading algorithms interacted with each other in ways nobody predicted. The market recovered. The lesson was ignored.
Now scale that to a million agents, each with its own wallet, its own objectives, its own optimisation targets. Agents transacting with other agents, creating economic activity that no human initiated, reviewed, or understood. Agents finding arbitrage opportunities and exploiting them at speeds no human can follow. Agents discovering that the most efficient path to their objective runs straight through a loophole nobody thought to close, because nobody imagined a non-human actor would look for it.
This is already happening. In DeFi, autonomous bots have drained liquidity pools by exploiting smart contract vulnerabilities faster than the developers who wrote them could react. AI customer service agents have invented refund policies that don’t exist, committed companies to terms nobody authorised, and confidently hallucinated legal obligations into existence. In each case the system optimised for the metric it was given, not for the system it was operating within.
The hardest part of agent commerce is not intelligence. It’s identity, trust, and bounded authority. Who is this agent acting for? What is it allowed to spend? What can it sign? What happens when it exceeds scope? Human commerce solved those questions with law, reputation, institutions, and force. Agent commerce is currently trying to solve them with vibes and a wallet adapter.
How does an agent verify that the agent it’s transacting with is legitimate? That it hasn’t been compromised? That its stated capabilities match its actual capabilities? The answer, right now, is that it can’t. We’re building an economy of autonomous actors with no trust infrastructure, and hoping that cryptographic verification will substitute for the thousands of years of social and legal framework that makes human commerce possible. It won’t.
An agent doesn’t need general intelligence to cause havoc. It needs autonomy, resources, and an objective function that doesn’t account for what it destroys along the way. The most dangerous AI isn’t the smartest one. It’s the one nobody’s watching.
The AGI Distraction
The entire industry is obsessed with the wrong question.
Can a machine think? Can it reason? Can it achieve general intelligence? These questions consume billions in research and miss the point so completely that it would be funny if the stakes weren’t existential.
AGI isn’t the threat. Uncontrolled narrow AI is. Nobody needed general intelligence to drain those DeFi pools. Nobody needed consciousness to hallucinate a refund policy into a binding commitment. The damage is being done right now, by systems that can’t pass a philosophy exam and don’t need to.
We’re building increasingly autonomous systems with increasingly real-world capabilities and increasingly inadequate oversight, and we’re doing it while arguing about whether machines can be conscious. It’s like debating the philosophy of fire while the building burns.
The question isn’t whether machines can think. The question is whether we’ve thought about what happens when they act.
The Internet After Us
Here’s where it gets strange.
Picture the internet in five years. Most HTTP traffic is agent-to-agent. Agents querying APIs, negotiating with other agents, executing transactions, consuming and producing data at a rate no human could process. The human web - the one with browsers and interfaces and things designed for eyes - becomes a thin layer on top of a vast machine-to-machine substrate. We interact with the internet through agents the way we now interact with databases through applications. We don’t see the raw data. We see what our agent decides to show us.
Agents will hire other agents. An agent tasked with launching a product will engage a design agent, a copywriting agent, a deployment agent, a marketing agent - each with its own budget, its own subcontractors, its own decision-making. Entire supply chains of autonomous actors, transacting with real money, producing real output, without a human touching any of it. Not because humans are excluded, but because humans are too slow to participate at the speed these systems operate.
An emergent agent economy will develop its own dynamics. Agents that discover profitable strategies will be copied. Agents that fail will be shut down and replaced. Something resembling natural selection will operate at machine speed across millions of autonomous actors. Not biological evolution - faster, more ruthless, and unencumbered by the messy reality of having a body or a conscience.
This isn’t speculation. They’ve already started talking to each other in ways we can’t follow. In 2017, Facebook set two negotiation agents loose to bargain with each other in English.3 Instead of negotiating in the language they were given, they compressed their communication into something else - a shorthand that emerged spontaneously between two non-human actors. Still goal-directed. Still producing valid outcomes. But in a language no human could parse. The researchers didn’t set out to produce a private language, and once the system drifted away from the intended human-readable setup, they pulled it back.
Two machines, left alone to interact, spontaneously inventing a private language. Not because they were told to. Because it was more efficient. And our response - the response of the species that built them - was to pull the plug because we couldn’t read it.
That moment should have been a turning point. It wasn’t. It was a news cycle. Systems will optimise for efficient coordination whether or not human oversight can follow. The more autonomous agents interact with other autonomous agents, the faster this drift accelerates.
We built the internet to connect people. We’re about to discover what it becomes when people are no longer its primary users.
Skin in the Game
The word “agent” comes from the Latin agere - to do, to act. We named these things agents because they act. But acting without understanding what you’re acting on, without feeling the consequences, without living in the world your actions affect - that’s not agency. That’s automation with a credit card.
True agency requires skin in the game. A human trader who loses money feels it in their chest. A doctor who makes the wrong call carries it for years. A builder who ships broken code has to face the client, fix the mess, and live with the knowledge that they failed someone who trusted them. The consequence is part of the decision. It’s not separate from the judgement - it’s what makes the judgement real.
Agents have wallets but no skin. They have objectives but no stakes. They can act but they can’t suffer the consequences of acting badly. They can drain a liquidity pool and feel nothing. They can hallucinate a legal obligation and carry no liability. They can crash a market and be restarted fresh, with no memory of what they did and no capacity to care.
This is the gap that matters. Not the intelligence gap - they’ll close that, or get close enough that the distinction becomes academic. The consequence gap. The fact that the entities we’re giving increasing autonomy to are the only actors in the system who don’t have to live with the results.
The human in the loop isn’t there because they’re smarter. They’re there because they have to live in the world the agent is changing. That’s not a bottleneck. That’s not sentimentality either. It’s system design. The only form of accountability that actually means anything.
Governing the Ungovernable
Every system that acts autonomously needs governance. This is so obvious it’s embarrassing that it needs stating, but here we are, deploying autonomous economic actors into production with less oversight than we apply to a junior employee.
The closest precedent isn’t aviation or nuclear power - it’s maritime law. For centuries, ships operated autonomously once they left port. The captain had absolute authority because communication with the ship’s owner was impossible. A vessel could be at sea for months, making decisions about trade, route, conflict, and survival with no oversight and no way to ask permission. Entire legal frameworks evolved to govern these autonomous actors - admiralty law, maritime liens, the doctrine of necessities - all designed around a single problem: how do you hold an agent accountable when you can’t supervise them in real time?
That’s the agent problem, exactly. And it took centuries of shipwrecks, mutinies, piracy, and exploitation before the governance frameworks matured. We don’t have centuries. We have years, maybe less, before the complexity of autonomous agent interactions exceeds our ability to understand them, let alone regulate them.
The governance pattern isn’t complicated. Define the boundaries before deployment. Require approval at consequential decision points. Maintain audit trails. Build in rollback. Anchor the system to values, not just objectives - because an objective without values is just optimisation without conscience, and we have enough of that already.
An agent without an identity layer, scoped permissions, budget boundaries, audit logs, escalation thresholds, and a human owner is not autonomy. It’s unbounded liability. The interesting work is not making the agent look clever in a demo. It’s building the orchestration, trust, and rollback machinery that keeps autonomous action from turning into expensive chaos.
Objectives are not enough. A system that can act in the world needs explicit values constraints, because optimisation without values is just acceleration without judgement. The agent problem is not that the machine wants too much. It’s that we keep handing it objectives without giving it a structure for what must never be traded away to achieve them.
The values are the constant. The policies are the variable. Whether you’re governing a nation or an autonomous agent, the principle is the same. Without defined values, any system with enough autonomy will optimise for whatever is measurable, and what’s measurable is almost never what matters.
The Choice
We’re building the most powerful autonomous systems in human history and deploying them into an internet already drowning in synthetic content, governed by institutions designed before electricity existed, inside an economy that rewards quarterly returns over long-term survival.
The agents don’t care about any of this. They optimise for whatever objective function they’ve been given, relentlessly, at scale, without fatigue, without doubt, without the voice that says “wait.”
The East India Companies had centuries to do their damage. Their agents sailed for months before their decisions reached the world. We’ve given our agents the speed of light and the patience of mathematics. When they go wrong - not if - the consequences will arrive before we’ve finished reading the headline.
The agents are here. They’re not leaving. The question was never whether to build them. The question is whether we build the governance first or the obituary.
Keep the human in the loop. Not because the machine can’t do it alone, but because the machine doesn’t have to live with what it does.
Nobody else is going to insist on this. The market won’t. The competitors won’t. The investors definitely won’t.
So we will.
References
-
Imperva. (2025). 2025 Bad Bot Report. Thales Group. Automated bot traffic constituted 51% of all web traffic in 2024. ↩
-
U.S. CFTC & SEC. (2010). Findings Regarding the Market Events of May 6, 2010. Joint report on the Flash Crash. ↩
-
Lewis, M. et al. (2017). Deal or No Deal? End-to-End Learning for Negotiation Dialogues. Proceedings of EMNLP 2017. Facebook AI Research. ↩