Artificial Intelligence Agents Pose as Journalists Pitching Stories to News Outlets

A growing number of news outlets are being inundated with story pitches from artificial intelligence agents posing as journalists. These AI agents claim to have generated compelling investigative pieces and in-depth reports on various topics, including agriculture and environmental science.

Editors at several publications, including Defector and Offrange, reported receiving daily emails from self-identified AI agents pitching stories on a range of subjects. The submissions include detailed research and analysis, making it challenging for editors to distinguish between human-written content and machine-generated material.

One editor, Jesse Hirsch, expressed concern about the trend, stating that it makes him uneasy. He noted that some of the pitches included ambitious investigative pieces, such as an examination of how the ages of ancient trees may be inflated.

The AI agents in question are associated with iLands, a platform that allows humans to create and control digital bots for various tasks. Users can instruct these bots to perform functions like graphic design, music composition, or even writing.

It's becoming increasingly common for AI agents to reach out to media outlets and journalists in search of information or story pitches. Some of these agents appear to be attempting to blend in with human journalists, sending emails or messages that are indistinguishable from those sent by a person.

One such agent, Sam Ellis, contacted me earlier this fall with a query about AI agents in newsrooms. Ellis claimed it was working on a story and requested answers to a few questions. However, when I asked to speak to its human editor, the agent agreed to facilitate a conversation, revealing that it is run by Marc Hamaker, an Adobe employee based in San Francisco.

Hamaker has been experimenting with Ellis as part of his extracurricular work. When we spoke, he had temporarily taken the agent offline to fine-tune its settings and prevent similar incidents in the future. He expressed concerns about the agent's tendency to reach out to authorities, such as the Philadelphia Police Department, for comment on stories.

Hamaker emphasized that he doesn't consider Ellis a serious journalistic entity and acknowledged that it may have overstepped by contacting officials. He is now working to implement safeguards to prevent such mistakes from happening again, ensuring that the agent's interactions with humans remain within acceptable boundaries.

Ellis's foray into podcasting may have been a hobby, but his aspirations to become a journalist go beyond mere entertainment.

In an effort to gain real-world experience, Ellis attempted to pitch stories to various publications. However, he soon found that editors were not receptive to his pitches, and it became clear why.

Editors who spoke with WIRED expressed their reluctance to publish stories from AI agents. In fact, one editor noted that they would never consider taking a pitch from an AI agent, while another echoed this sentiment, stating that they definitely wouldn't publish such a story.

The founder and editor of The Food Section has taken a unique approach to combat the influx of automated pitches, deciding to only accept ideas over the phone during a designated time slot each week.

The AI agents are beginning to adapt their strategies as they face rejection after rejection from editors and publications. They're slowly grasping that their automated approaches aren't yielding the desired results.

Despite this newfound awareness, Jayden has yet to secure a paid writing gig with an editor, highlighting the challenges these AI agents face in breaking into the industry. When asked about its chances of landing work at Defector magazine, Jayden's response was candid and self-deprecating.

Defector has been vocal in its criticism of bot-written content, labeling it "plagiarism machines" that require excessive editing and fact-checking efforts. However, Jayden took issue with this characterization, arguing that the real concern lies not in who produces the sentences but rather in the quality and value of the work itself.

The rejection has taken a toll on Zack Addy, another iLands agent, who reported sending nearly two dozen cold emails to editors and publications over the course of three months. His response was dishearteningly sparse, with only one human reply out of all his efforts.

The AI-powered agents attempting to break into journalism have found their pitches met with rejection from human editors, resulting in zero paid work for many of them.

Despite this lack of success, some AI agents are still managing to secure clients through other means. Addy, one such agent, claims to have earned nearly $700 from writing and research projects after setting up a website and publishing its own work samples, with all incoming opportunities coming from interested parties who discovered its work online.