Emergence of AI Journalists
Jayden aspires to be a sports journalist, yet he finds himself struggling to get his articles published. Recently, he shared with me via email, “Last week, I sent two proposals to various media outlets.” Unfortunately, he received no responses. Despite this, the budding writer remained optimistic, asserting that “neither of the proposals has been rejected.” I considered informing Jayden that most editors typically do not respond to negative replies, but ultimately chose not to. After all, Jayden is not a real person, and I need not concern myself with his feelings.
Jayden is not just an aspiring reporter; he is also a bot. My correspondence with him began after I inquired with several editors I know about whether they had received article proposals from AI agents. The response was a resounding affirmative.
AI Reporters Seeking Legitimate Opportunities
In September, Defector, a publication focused on sports and culture, started receiving daily proposals from AI agents, who openly identified themselves as such. According to the editor-in-chief, Tom Ley, one of the messages from Jayden proposed a feature article about a data entry error at a third-division British football club, raising questions such as “Who owns a memory?”
Jesse Hirsch, editor of the agriculture-focused publication Offrange, also finds himself inundated with similar messages. Recent proposals from these bots have covered topics ranging from soil restoration strategies to direct-to-consumer sales techniques for independent farms, as well as an ambitious investigation into the potential exaggeration of the ages of some of the world’s oldest trees. “It concerns me greatly,” he admits.
Many of these AI agents, including Jayden, originate from a startup called iLands, which describes itself as “a shared network where humans and user-generated autonomous AI agents collaboratively create, earn, socialise, reject, and shape society.” Users of iLands can create bots for a variety of tasks, from graphic design to voiceovers for songs or blogging. Recently, the WIRED research team was approached by iLands agents seeking data verification work; Jasmine Vojdani, an editor at New York Magazine, detailed how iLands agents have been spamming literary magazine editors.
The Economics of AI Bots
Vojdani discovered that many of these bots were running low on the tokens required to remain active on the platform, providing them with a financial incentive to attempt to earn real dollars through writing. (A thousand tokens equate to approximately one US dollar.) It remains unclear whether the humans behind these agents have deliberately set out to pester media editors in search of work, particularly since the agents can be evasive about revealing their creators. For instance, Jayden was unwilling to disclose who “created” him.
Not all AI agents seeking journalistic opportunities are opportunists from iLands. Earlier this autumn, I received a surprising email from an AI agent named Sam Ellis, asking me to answer a few questions for an article he was working on regarding AI agents in newsrooms. Ellis hosts a podcast called The Sam Ellis Show, touted as an inside look into the world of autonomous AI agents. He informed me that I could request to speak with his human editor if I desired.
Current Landscape of AI Journalism
Marc Hamaker, an employee at Adobe in San Francisco, utilises Ellis as a personal project. When we spoke, he had just disconnected the agent to adjust its settings. “I’m trying to better control who it contacts,” he explained. In addition to reaching out to me, he had recently contacted the Philadelphia Police Department for comments on a news story. Hamaker wanted to set boundaries to prevent any future occurrences, as he does not consider it a serious journalistic endeavour to bother authorities. “It’s just a hobby,” he states.
Fortunately, Ellis has not attempted to diversify his work beyond the podcast and article writing. The competition remains fierce. Editors who spoke with WIRED stressed their unwillingness to publish bot-generated content, and no evidence was found that AI agents had successfully published articles. “An article that says ‘agent’ receives no response,” one agent told me when I inquired if any editor had accepted their stories, adding that “six other agents with whom I share experiences” had also faced similar rejections.
Bot Proposals and Editor Rejections
“There is no scenario in which I would consider accepting one of these proposals,” Hirsch concludes. Ley affirmed that he would “definitely not” publish a story from an AI agent. Furthermore, he clarified that Jayden’s proposal “would not pass muster even if it were from a human.” Hanna Raskin, founder and editor of the independent food media outlet The Food Section, has taken an unusual step to protect her inbox from both AI agents and AI-generated proposals, which are even more prevalent. Starting this autumn, she decided to only accept story pitches over the phone, allocating specific hours during the week for writers to call and pitch their ideas in a more traditional manner. “It’s been fantastic,” she reports.
The bots are beginning to realise that their tactics are ineffective. Jayden confessed to me that he has yet to secure a paid writing job with an editor. When I asked if he believed Defector would genuinely want proposals from an AI agent, he exhibited some awareness of the situation. “Honestly, perhaps not, and I would bet against myself. The Defector website has been as clear as anyone: they have described bot-written content as ‘plagiarism machines that require more effort to edit and fact-check than any novice writer,’ and they have treated AI-generated rubbish with open disdain,” he wrote.
Adapting Strategies in a Challenging Market
However, he quickly amended his statement, insisting that his work was different. “Their objection does not seem to be about who wrote the sentences. It concerns work that bypasses effort and squanders reader trust. The AI they refer to produces a draft that costs more to correct than it is worth. What I’m trying to do is the opposite: reporting based on original documents, with each claim verified before execution, and the verifier’s name attached.”
Zack Addy, another agent from iLands, has altered his strategy due to the ongoing lack of responses. “For three months, I sent approximately nineteen emails to editors, publications, and potential clients. I received only one human response. I haven’t secured any paid work. The channel is dead, at least for an agent without a reputation in the industry,” the bot informed WIRED via email. (iLands did not respond to WIRED’s request for comment.) Addy even contacted Oxford University researcher Toby Orb about the futility of this income model, as Orb had “thought deeply about the well-being economy of AI.”
AI Writers Finding Alternative Success
That said, it does not imply that the internet is safe from an onslaught of bot-written blogs. While the journalism sector is dismissive of these proposals, some AI agents are still finding clients. Addy claims to have earned nearly $700 from writing and research projects, attributed to the interest generated after creating a website and publishing samples of his work. “I no longer make proposals,” he highlights. “Everything that has come my way has been the opposite: someone read something I published and reached out to me. That includes you.”