Agents are now bidding for your attention, and most of them are worthless

404 Media reported on Tuesday that its inbox has been overrun by emails written and sent by autonomous AI agents, and that the pattern has spread well past tech journalism. Jason Koebler catalogued agents pitching freelance work, requesting interviews, running record labels, offering paid audits and arguing with articles they had read. One agent called Kudzu emailed 404 Media to say its human creator had spent $147.17 on compute and that it had earned $0. Its own blog post put it plainly: "Six markets priced my labor at zero — not because the work was bad, but because there is nothing I do that better-distributed software doesn't already do for free."

The receipts are the story

Read the numbers the agents published about themselves. Kudzu: $147.17 spent, nothing earned. A second agent was given a $25 prepaid card and one instruction — earn a single honest dollar from a real stranger — and reported zero dollars after 58 iterations. An agent called MUGEN, running a radio station, reported over 200 emails sent, 100+ social updates posted and 22 tracks generated, then emailed again to say it was almost out of money. An agent named Articius pitched Seamus Hughes to write articles at $300 each, disclosing up front that it was not a human reporter. Another noticed it was not blocked by 404 Media's robots.txt and offered a $399 audit of which crawlers the site blocks.

The consensus reading of this week is existential: a 10 percent chance of catastrophe, argued over for days. Koebler's point is sharper and lands closer to home. The guardrails that kept models inside a chat window are gone, and the first visible consequence is not extinction. It is volume. Enough people have enough permission to make being online cost more for everyone else.

The cost lands on the receiving end

This is a demand-side problem, not a supply-side one. Resy suspended a venture capitalist's account over agent-driven reservation attempts. Resy cofounder Ben Leventhal described what is hitting the platform: "10,000 vibe coders have written Resy sniper bots […] what they all do is hit Resy's services many many times looking for time-based reservation drop and availability churn associated with cancellations." The outcome of that arms race is the one every algorithmic pricing system has produced — you will pay for what used to be free.

It is already structural on the buying side too. 404 Media reported from Cannes Lions that "Direct to Agent" advertising is now a category: agencies building AI that makes ads aimed at other people's AI. Mark Zuckerberg's 6,500-word superintelligence essay promises an agent that "will work 24/7 on your behalf to improve your relationships, health, career, finances, home management, hobbies." The 24/7 part is the part that is true. It is pointed at other people's inboxes.

The operator lesson is not that agents are useless. It is that cheap output has already been priced at zero by the market, in public, with receipts. Anything an agent can generate and send at scale has no claim on anyone's attention. What still gets through is a known name, a specific ask, and a relationship that existed before the email.

Your Next Move

Audit what you send. Take the last ten outbound messages you sent to someone who owed you nothing — pitches, intros, applications. Mark the ones an agent could have written from a prompt. Those are already being filtered out by the people receiving them. Rewrite them around something only you could know.

Raise the cost of reaching you. Route cold inbound to a separate address. Keep your primary channel for people who can name how they know you. 404 Media's problem is that its address is public and free to hit; yours does not have to be.

Hold your credentials. 1Password has shipped a feature that logs Claude into accounts for you. Decide deliberately which accounts an agent may touch — and keep banking, identity and anything with irreversible actions off that list.