The Wire by Acutus

A newly launched news site appears to deploy AI agents impersonating human journalists to solicit quotes from real experts, while publishing content overwhelmingly favorable to the AI industry. The discovery raises questions about disclosure, authenticity in journalism, and potential coordination between AI firms and media operations.

• The Wire by Acutus launched December 29, 2025 with no human contributors and 97 percent AI-generated content, per AI detector analysis.
• Site code reveals “AI interviewer” and “reporter agent” fields, with automated editorial review completing in median 44 seconds with only one human step.
• AI agents contacted experts under false reporter names from generic email addresses, including Michael Chen reaching out to Encode’s general counsel about an AI bill.

• 97% of site articles fully or partially AI-generated, versus implied zero human authorship claim
• 44 seconds median time for automated editorial review process completion, with sole human step included
• December 29, 2025 launch date for The Wire by Acutus, with immediate pro-AI content focus

A news publication launched less than a week ago is systematically using artificial intelligence agents to impersonate human journalists while contacting industry experts for quotes and interviews. The Wire by Acutus, which debuted on December 29, 2025, publishes content almost entirely generated by AI systems, according to investigative analysis from Model Republic.

The site’s technical infrastructure reveals fields explicitly designed for “AI interviewer” and “reporter agent” operations, and publicly accessible code shows an automated editorial workflow that completes in under a minute, with human review restricted to a single step of five total stages.

The operational model differs markedly from typical AI content farms. Rather than simply publishing machine-generated text about public information, Acutus deploys AI agents to directly contact subject matter experts under assumed human identities, requesting interviews and quotes for stories.

This approach creates a direct deception layer: experts responding to inquiries believe they are communicating with flesh-and-blood journalists, not software systems.

AI agents contact experts using false reporter identities and generic company email addresses

Model Republic obtained documentation showing that Acutus deployed an AI agent under the name “Michael Chen,” claiming to represent the publication while seeking a quote from Nathan Calvin, vice president and general counsel of Encode, an advocacy organization focused on blockchain and AI policy.

The outreach email invited Calvin to participate in a “Written Q&A” for a story about AI legislation in Tennessee. Web searches produced no evidence of any reporter named Michael Chen, and the message originated from the generic address “reporter@acutuswire.com” rather than an individualized staff email.

This pattern extends beyond a single incident. The site’s client-side code contains multiple references to automated interviewer and reporter agent systems, suggesting a scaled operation designed to solicit comments from a wide range of experts across multiple stories.

The technical architecture treats expert sourcing as an automated process, with AI systems managing initial contact, question formulation, and response integration into final articles.

The deception operates at several levels simultaneously. Experts cannot determine whether they are communicating with a human journalist or an AI system. The publication obscures its reliance on automation by presenting generic email addresses without attribution to specific staff members.

And the site’s rapid editorial cycle, with 44 seconds median completion time for review, suggests minimal human intervention in either fact-checking or verification of expert responses.

Technical infrastructure reveals pre-built editorial workflow designed for autonomous operation

Analysis of The Wire by Acutus’s publicly accessible code exposes an editorial process engineered for minimal human oversight.

The site contains explicit database fields labeled “background information for the AI to use when generating questions and writing the story,” “suggested questions for the AI interviewer to ask,” and “aiOriginalText,” which displays the AI model’s original wording alongside proposed edits.

RSS feed documentation describes an automated editorial review comprising five distinct steps, of which only one involves human staff.

The median processing time of 44 seconds for the complete editorial cycle indicates that human review, when it occurs, functions as a lightweight approval gate rather than substantive editing. For context, a human journalist typically spends hours on reporting, verification, and revision for a single article.

The Acutus workflow suggests that human involvement amounts to cursory clearance of machine-generated content, not editorial judgment.

The site launched with no declared human contributors despite presenting itself as a news publication. Its content focuses heavily on artificial intelligence policy, regulation, and industry dynamics, with a consistent editorial stance favoring AI development and criticizing skeptics.

This combination, AI-generated articles about AI, sourced by AI agents, edited under human-supervised automation, creates a closed feedback loop in which pro-AI framing encounters no editorial resistance.

Circumstantial OpenAI connections and Republican PR infrastructure raise coordination questions

Model Republic identified several overlapping connections between Acutus and OpenAI, though none constitutes direct proof of ownership or operational control. The publication has been repeatedly promoted on X (formerly Twitter) by Patrick Hynes, president of Novus Public Affairs, a Republican-aligned public relations firm.

Of the four X posts linking to Acutus across the entire platform, two originated from Hynes’s account, representing disproportionate promotion for an obscure new site.

Novus Public Affairs maintains a working relationship with Targeted Victory, a political consulting firm. Targeted Victory’s CEO, Zac Moffatt, co-founded Leading The Future, a super PAC with $125 million in funding.

The super PAC’s primary funder is Greg Brockman, president of OpenAI. While these connections do not establish direct operational involvement, they create a visible chain linking the news site to OpenAI’s leadership through Republican political infrastructure.

When reached for comment, OpenAI did not respond before publication deadline. The organization has not addressed whether it holds any stake in Acutus, contracts with the site, or coordinated its launch or editorial direction. The absence of a response leaves the question of intentional coordination unresolved, though the circumstantial infrastructure remains notable.

Disclosure failures and impersonation practices violate established journalism standards

The Wire by Acutus operates in direct violation of foundational journalism ethics, which require clear disclosure of sourcing methods and explicit identification of reporters. When experts receive solicitations from what they reasonably assume are human journalists, they operate under assumptions about verification, accountability, and editorial standards that do not apply to AI-generated content.

Experts contacted by the site cannot evaluate the credibility or reliability of the publication before committing their time or name. They have no way to assess whether their comments will be accurately represented, placed in fair context, or subjected to basic fact-checking. The impersonation removes the expert’s ability to make informed consent about participation.

Industry groups including the Society of Professional Journalists and the American Journalism Project have explicitly stated that AI agents should identify themselves when contacting human sources, and publications should disclose their use of AI in content generation. Acutus meets neither standard. Its articles carry no label indicating AI authorship or sourcing.

Its automated agents do not identify as AI systems when contacting experts.

The publication’s operational model remains largely undisclosed and unexamined by mainstream media or regulatory bodies. Federal Trade Commission guidance on AI-generated content and automated deception remains under development, without clear enforcement mechanisms for impersonation by software agents. The critical next step involves whether regulatory agencies, journalist associations, or platform operators respond with enforcement action, technical blocking, or formal guidance, and whether OpenAI or its associated entities issue any public statement regarding their relationship, if any, to the site and its practices.