Ten teams have been selected as finalists in this year’s Student Innovation Competition from the Reynolds Journalism Institute. Over the course of the fall semester, students from colleges and universities across the country will compete to develop merchandise items that can help community news organizations raise funds or awareness. The competition is part of broader efforts at RJI…

Source

Are you a journalist who has used AI for work in some capacity but been nervous to tell anyone about it? Worried about how your bosses might see you, or that your work might be discredited?

Well, academics are in the same boat.

A new study, titled “Guidance over guidelines? Unpacking the uses and concerns of generative AI in communication science” and published in Information, Communication and Society, uncovers how communication scholars are using generative artificial intelligence in their academic work.

Researchers from the National Yang Ming Chiao Tung University in Taiwan, the University of Amsterdam, and Utrecht University surveyed more than 1,100 communication scholars from 77 countries about their use of AI, their concerns about AI in the workplace, and the kind of guidance they seek about AI use in their work. They also held four focus group discussions.

: ?

This is the first systematic, field-wide, international empirical investigation of generative AI practices in communication research. We surveyed 1,138 scholars across 77 countries + did 4 focus groups. Small thread with 5 key findings

[image or embed]

— Hannes Cools (@hannescools.bsky.social) September 28, 2026 at 6:50 AM

The study reveals a disconnect between researchers’ beliefs about AI and their use of it. Nearly 70% of survey respondents already used generative AI in at least one research-related task, but respondents also mostly considered AI use to be inappropriate for most research-related tasks. While just over half of respondents said that using AI for writing texts (55%) and generating code (50.4%) were appropriate use cases, the study’s authors note that survey participants did not reach a consensus on appropriate or ethical uses of AI in their work.

The lack of institutional guidance on how best to use AI in research means researchers must develop their own ethical standards. This creates a stigma around AI use, according to the study’s authors. Refraining from AI use may put researchers at a disadvantage and slow down their work, but using AI could also be a disadvantage if journals reject research that discloses AI use. “Researchers may recognize an ethical obligation to disclose genAI assistance while fearing that doing so will lead others to perceive their work as less authentic or credible,” the study’s authors write. “This tension has been described as a transparency paradox.”

All of this results in feelings of “AI shame,” the authors say.

“While such hesitation is understandable given that reporting genAI use might reduce the chances of publication, it ultimately hampers open discussion about how these tools can be used ethically and responsibly in research,” the study says. “Transparency is therefore essential, as effective and meaningful guidelines can only be developed when we understand when and how genAI is being used.”

The study’s authors recommend that the communication research community — including individuals, institutions, and journals — create a “disclosure-positive” culture in its field by requiring researchers to disclose the stages of their work at which they used generative AI.

“Simply forbidding the use of genAI — or pretending that it is not being used — is neither sustainable nor constructive,” the authors write. “Instead, acknowledging its presence and defining transparent boundaries for its use will help align ethical standards with actual scholarly practices, fostering honesty, accountability, and trust within the research community.”

Read the full study here.

Artificial intelligence is making editors do strange things, like second-guess skills built over entire careers or blanch at the sight of punctuation marks. This manifested at the recent ACES: The […]

The post AI learned how to write from humans. Now humans are trying not to sound like AI. appeared first on Poynter.

OpenAI announced on Monday that it’s doubling its support for the Lenfest Institute’s AI Collaborative and Fellowship Program. Since the program launched in October 2024, it has placed AI fellows in 11 local newsrooms across the U.S., mainly in metro newspapers like The Philadelphia Inquirer, The Baltimore Banner, and The Minnesota Star Tribune.

As the program moves forward, Lenfest plans to keep the same approach of embedding full-time engineering staff in local newsrooms for two years. In this next grant cycle, though, the program is expanding its reach beyond metro newspapers to consider fellowships for local TV news and statewide public service journalism.

“The key success factor seems to be that the fellows are hired locally, report locally, and are viewed as part of their local team, because they are,” Jim Friedlich, executive director and CEO of the Lenfest Institute, told me. “They’re not Lenfest employees, and they’re certainly not Microsoft or OpenAI employees. They’re free to work in any technology that they deem appropriate, and in every case, they were the first full-time AI employee at each news organization.”

The first cohort was funded with $10 million from OpenAI and Microsoft. Each company contributed $2.5 million in cash and $2.5 million in credits that could be used for enterprise licenses, cloud storage, and tokens for generative AI products.

Microsoft began distributing its grant funds six months after OpenAI and hasn’t decided yet whether it will renew its commitment once that cycle ends in 2027. OpenAI, meanwhile, is expanding its support, contributing $10 million to the program on its own: $5 million in cash and $5 million in credits.

“Funding supports people and implementation; credits allow for experimentation and make our technology accessible,” said Tom Rubin, the chief of intellectual property and content at OpenAI. “The goal is to create lasting capability for news organizations: deeper research, stronger reporting, richer user experiences, improved workflows, and more sustainable businesses.”

When the fellowship launched in 2024, OpenAI had just closed a funding round that valued it at $157 billion. Earlier this month, The New York Times reported that the company is considering a new funding round that could value the company at $1.5 trillion ahead of its IPO.

Alongside that growth, the company is expanding its role as a funder of journalism initiatives and institutions. In July, OpenAI renewed its funding of the Product & AI Studio, a program run through The American Journalism Project (AJP) to help nonprofit local newsrooms experiment with AI. The $5 million in cash and $3 million in credits will support the more than 50 news organizations in AJP’s portfolio. They’re encouraged to build tools and infrastructure that can be used by other newsrooms.

This month, OpenAI also announced it is donating subscriptions for more than 400 enterprise-level ChatGPT accounts to students and faculty at the Medill School of Journalism at Northwestern University and the Craig Newmark Graduate School of Journalism at the City University of New York. In the announcement, OpenAI hinted at further donations to journalism schools down the line.

The Lenfest Institute fellowship is unique in that it brings full-time staffers with technical expertise into news organizations and assigns them newsroom leaders, or “executive sponsors,” to manage their work and ensure they aren’t isolated.

“You get a full-time person to follow the thread and be creative and attack problems as they come in,” said Tristan Loper, the head of national programs at Lenfest. “It’s not a part-time, or consulting, or project-based role. You live and breathe the problem…I think that leads to a lot of organic problem solving and cross-team collaboration.”

As part of the fellowship, participating newsrooms join collaborative Slacks, travel to fellowship summits with training from OpenAI and Microsoft, and receive support to attend journalism conferences. After The Inquirer presented Dewey to the rest of its cohort, a fellow from The Baltimore Banner said they had been working on similar problems with archives in their own newsroom.

“Their fellow sat down with Kevin, walked through the code, and made some adjustments to their particular setup,” said Boggie. The Banner has since launched its own version of the tool.

A fellow from Chicago Public Media, which owns WBEZ Chicago and the Chicago Sun-Times, also customized a version of Dewey to interact with radio transcripts. Dewey is open-source, and the underlying code is available on GitHub to newsrooms outside the program as well.

Moving into the second iteration of the fellowship, Lenfest wants to emphasize the collaborative parts of the program further, prioritizing reusable AI tools that can be easily copied and customized for other news organizations.

As part of that push, Lenfest has brought on Sonali Verma, who leads INMA’s Newsroom Innovation Initiative and previously worked on Sophi, an AI-powered paywall product first built in-house at Canada’s Globe and Mail. This week, Lenfest is launching a new newsletter called Context Window, written by Verma, that will share what fellows have learned and link to specific tools and GitHub resources. It’s “all with an eye toward how this can be adopted” by other newsrooms, Loper said.

Chase Davis, formerly at The Minnesota Star Tribune’s AI Lab, where he oversaw the Lenfest fellowship, has also come on board as the program’s technical lead. Among other responsibilities, he’ll help newsrooms productize any AI-powered tools they develop.

Lenfest is encouraging all 11 of the current participating newsrooms to reapply for the next grant cycle, but Friedlich said half, at most, will be reaccepted, leaving room for more news organizations to join.

With this next cohort, Friedlich says he’s interested in exploring how AI can support the business of local TV news. He also sees promise in using AI to support regional public service journalism, such as accountability reporting on state legislatures.

OpenAI’s renewal comes as the U.S. local news industry continues to contract, in part because of traffic declines related to AI search. This month McClatchy laid off more than 90 unionized workers across 70 publications. USA Today Co. also made major cuts, referencing AI search–related traffic declines. Against this backdrop, many local newspapers are turning to AI to assist with story drafting, while AI agent–led news publications with little to no human editorial oversight continue to pop up across the country.

Lenfest says it will not invest in newsrooms that plan to use AI primarily to displace human reporting and writing.

“We believe really strongly that there should be a human in the lead, not just a human in the loop — [someone who] is leading the use of AI and its personalization and its adaptations, rather than double-checking the work when it’s about to go out,” said Friedlich. “That may sound old-fashioned, but we think it’s really important.”

This story has been updated to correct Chase Davis’s role at The Minnesota Star Tribune.

Once you’ve identified the impact signals you want to spark, think about opportunities to do so beyond your coverage by designing events, programs and spaces that help you both create and learn from your newsroom’s impact. This learning loop can guide decisions about where to invest your time, reporting and social capital next.

Impact-minded organizations build habits that help them notice what their communities are responding to, interpret what those signals might mean and adapt their journalism while the work is still unfolding.

Ambassador networks, volunteer pathways like Documenters, community advisory groups or even dismantling beat structures in favor of communities of interest — these are systems for learning as much as they are engagement strategies.

How one newsroom built a learning system with community ambassadors

Rather than relying solely on audience metrics, Rocky Mountain Public Media is experimenting with a statewide network of community ambassadors to understand emerging issues, community priorities and patterns of participation that help inform future editorial decisions.

Amber Coté, senior director of statewide civic engagement at RMPM, writes about how they track qualitative and quantitative signals of impact through their community ambassadors, identifying opportunities to further engage and documenting the stories behind data that might look small on a spreadsheet but signal important shifts.

Ultimately, the purpose of impact data, and of reading signals of impact more consistently, isn’t to prove our role as stewards of news and information. The purpose is in understanding how we create value so we can create more of it.

Try it out: Measure impact for the duration of a project

Start here: Choose one project that’s underway. It could be an investigation, a video series, a live event series or even a news product like your mobile app, a newsletter or an interactive database. List out the impact goals you may have identified during last week’s impact meeting exercise.

Take action: Make a plan to assess the metrics and engagement of that project on an ongoing basis. Determine:

• Which metrics or signals align with your impact goals?
• How often will you review the progress toward your goals?
• How will you pivot if something isn’t landing?

Reflect: Do you have access to the data you need to continuously track impact? If not, here’s one way to start.

Dig deeper:

• Growing a robust public network for better public media
• Design events and digital spaces to create — and measure — impact
• We built civic action boxes for readers. They became our impact data.
• Turn impact into better decisions

The post Create an impact learning loop appeared first on American Press Institute.

It has been a troubling couple of weeks in the media as President Donald Trump, in perhaps his most brazen act as president, banned prominent news outlets from covering his […]

The post Rachel Maddow on why Trump keeps attacking the media: ‘You play the hits’ appeared first on Poynter.

My friends and family know I’m a fact-checker, so they send memes on a regular basis, asking me, “Is this true?” It’s usually a post that makes them sad or […]

The post Fact-checking isn’t just for journalists. Here are 5 reasons it’s for you, too appeared first on Poynter.

As the Trump administration openly defies a federal court order restoring access for CNN, MS Now, and Politico, the White House press corps needs to hold its nerve.