Artificial intelligence has helped Editor & Publisher complete tasks in minutes that once took a day and build the capacity to support a doubling of its advertiser count in one year. But as AI became more embedded in the business, Publisher Mike Blinder wanted greater confidence that the organization was using it responsibly—and could demonstrate that to others.
That decision led E&P to become the first publisher to earn Ethical AI Certification from the Alliance for Audited Media.
In a September 10 E&P Solutions Series broadcast sponsored by AAM, Blinder, E&P Editor-in-Chief Gretchen Peck and AAM CEO, President and Managing Director Richard Murphy took viewers inside the process. Their discussion explored what certification examines, what changed at E&P and why establishing clear rules can give publishers greater confidence to innovate.
For Blinder, AI’s business benefits connect directly to E&P’s ability to support journalism. Greater efficiency allows the company to develop advertising solutions, serve more partners and invest in reporting on the industry. “We’ve doubled our advertiser count, and AI has given us the capacity to support that growth,” he said.
That expanding role also made accountability more important. Blinder described E&P’s relationships with readers and advertisers as relationships in which trust must be earned continuously. Independent review offered a way to examine practices that had developed as the team experimented with new tools. “Opening those practices to independent review kind of made me sleep better at night,” he said.
E&P’s position as an industry publication added another responsibility: demonstrating the standards it encourages other publishers to consider. “We believe we have the responsibility to lead by example,” Blinder said. “That meant opening our policies and practices to independent scrutiny.”
Murphy said the certification originated with publishers seeking help addressing audience concerns about AI. They also wanted a way to distinguish their work from the growing volume of low-quality, automatically generated content.
As AAM introduced the program, it discovered that many organizations needed help building their approach before they could seek verification. “We found out that publishers really need education and a roadmap on how to get there,” Murphy said.
AAM distilled extensive guidance into eight pillars addressing governance, transparency, rights and permissions, human oversight, bias, data protection, employee education and risk management.
The distinction between a written policy and an operating practice is central to the review. Murphy invoked the auditing principle of “substance over form”: publishers must demonstrate that they implement and follow their policies. He described the certification process as “thorough but not arduous.” When reviewers identify a gap, they explain it, recommend ways to address it, allow time for changes and check again.
Peck led E&P through virtual meetings with AAM’s audit team, working through the pillars and discussing how the organization applied its policies.
E&P already had a public-facing AI policy. Certification prompted refinements and more explicit guidance for the people doing the work. “I don’t think it necessarily changed our position on how we were going to use AI,” Peck said. The value came from being able to “clarify that and codify it for our team.”
One practical issue involved ongoing education. E&P relies heavily on freelance journalists, each with individual reporting processes and preferred tools. The review required the organization to consider how training would reach that distributed team.
E&P determined that it would monitor the industry for free or low-cost professional development opportunities and make them available to contributors.
The team also provided feedback on the internal policy. One contributor strongly discouraged AI-generated images accompanying their reporting, believing those images detracted from the journalism.
That concern illustrated why responsible adoption requires conversation with the people whose work and credibility are involved.
AAM's framework distinguishes between a general policy explaining an organization’s AI use and disclosures accompanying particular content.
Murphy explained that AAM’s program requires disclosure of synthetic images created through prompts. He distinguished that use from assistance with editing or polishing copy, which does not carry the same content-level disclosure requirement under the program.
E&P discussed its labeling of AI-generated imagery and how it would display the certification seal, including on its public policy page.
An audience member pushed the transparency discussion further by asking which platforms the participants actually used.
Blinder described using ChatGPT and Descript in business and production work. Peck discussed tools including ChatGPT, Gemini, Claude and transcription services used by contributors. Murphy noted that AI is increasingly embedded in everyday software and said AAM uses a secure Copilot environment for sensitive organizational information.
The exchange brought the discussion back to concrete practices: transparency becomes more useful when publishers explain where tools enter their work and how people oversee the results.
Blinder and Peck agreed that governance can make publishers more comfortable exploring AI. Clear expectations help teams understand how to proceed and when to exercise caution.
Murphy said publishers can continue using AI while developing their governance practices and working through improvements. Small organizations can participate, too; their direct oversight and ability to make changes quickly can be advantages.
His recommended starting point is AAM’s compliance guide, which explains the pillars and outlines the documentation publishers should prepare.
Certification also requires continued attention. Murphy said the program calls for policy review at least annually, with more frequent review when an organization’s AI use changes substantially.
Peck emphasized that E&P’s work will continue. “You have to approach this as an ongoing conversation with your newsroom team and throughout your news organization,” she said.
Blinder closed by connecting that ongoing work to the reason E&P pursued certification in the first place: “Trust cannot be added after an AI problem occurs.”
Comments
No comments on this item Please log in to comment by clicking here