Generative AI and journalism: a difficult relationship
This is the first of a series exploring GenAI and journalism
Journalism and the digital age had a complicated relationship since the beginning. For publishers worldwide, finding a singular path to adapt their traditional business models has proven elusive. The oligopoly of news production and distribution has shattered, disintermediation has intensified competition in content creation, and news organizations have struggled—and continue to struggle—to carve out a sustainable way forward. Advertising revenue alone is insufficient, platforms have monopolized audiences, and algorithmic challenges have added new layers of complexity.
On the other hand, the digital revolution enabled publishers to reach diverse audiences, experiment with new storytelling formats, and leverage data-driven strategies for greater impact. These advancements hold the promise of transforming journalism into a more inclusive and innovative space, but they came with a cost and, moreover, the audience is not granted anymore—assuming that it has been, once.
The rise of generative AI follows and scale this path, once again.
Production and the temptation of quantity and speed
Throughout the history of journalism, every opportunity to increase the speed and volume of production has been eagerly embraced—often at the expense of quality and production of original journalism. This relentless pursuit of output increased the usage of pre-packaged material such as press releases or stories provided by news agencies to produce content. There is a term for this: churnalism.
With the rise of the web, the strategy of competing on quantity and speed proved a near-instant failure, yet it was pursued vigorously for years. Publishers chased traffic from Google, then from social media (which quickly became walled gardens), and finally from any available source to feed the demands of an advertising-driven business model.
Generative AI seems poised to amplify this dynamic. By automating the production of articles, summaries, and content variations, AI enables unprecedented levels of scalability and it also proves that value is not in quantity or speed.
As history has shown, volume alone does not sustain journalism. Flooding audiences with mass-produced, low-value content erodes trust, dilutes brand identity, and fails to distinguish credible newsrooms from the deluge of noise on the internet. Many will undoubtedly continue to focus solely on quantity, unfortunately. It's already happening.
This relentless pursuit of quantity and speed has been a hallmark of journalism's adaptation to digital challenges, often at the expense of quality. Generative AI has the potential to either exacerbate these tendencies or offer solutions to elevate journalism—depending on how it’s used. However, the technology’s introduction has sparked significant resistance, not just from those wary of quality erosion, but also from legal and ethical perspectives.
Lawsuits and resistance
In Italy, and other conservative contexts, resistance to generative AI often stems from fears of displacement. Journalists, particularly those working in established newsrooms, worry about losing editorial control or compromising quality. Even tools designed to save time and streamline repetitive tasks—like AI for transcription or headline suggestions—face rejection.
This resistance contrasts sharply with the more pragmatic approach seen in countries where publishers actively develop proprietary tools or make experiments.
While some fear that adopting AI could undermine journalistic integrity or lead to job losses, others have pushed back through the courts, defending intellectual property and ethical standards against what they see as AI overreach. These lawsuits reveal the tensions at the heart of journalism's evolving relationship with AI.
On the legal front, resistance has manifested in a series of high-profile lawsuits, highlighting concerns over copyright infringement, data privacy, and misinformation. Below is an overview of the most significant cases.
Could major Silicon Valley firms legally scrape content to train Large Language Models (LLMs)? The answer remains ambiguous, hinging on ongoing lawsuits and forthcoming regulatory interpretations. In the European Union, the application of existing regulations to AI technologies is under scrutiny, with policymakers deliberating on frameworks to govern AI's use of copyrighted material.
The tension between publishers and AI companies isn’t limited to philosophical or professional concerns. At its core, much of the debate revolves around economics—specifically, who benefits financially from the content that fuels AI systems. While legal cases seek to establish boundaries, many publishers are striking deals with AI companies, balancing compensation with the promise of technological innovation. These agreements reveal another layer of the evolving relationship between journalism and generative AI.
Follow the money
The whole battle is probably in large part a financial issue: amidst this uncertainty, publishers and AI developers are proactively forging agreements.
These deals aim to balance the interests of content creators with the technological advancements of AI firms. Below is a comprehensive overview of notable agreements between publishers and AI companies.
Unfortunately, the vast majority of these agreement is opaque and the detailes have not been publicly disclosed.
Automation for a better journalism
Amidst the high-level negotiations and legal disputes, there remains a practical side to generative AI's impact on journalism. Beyond the financial and ethical debates, AI offers tangible opportunities to streamline workflows and enhance productivity. By automating specific tasks, journalists can focus on what they do best—investigating, contextualizing, and creating meaningful stories. Here are some of the most practical and impactful ways AI can assist in daily newsroom operations:
SEO-Optimized Headlines
Generative AI can create variations of headlines tailored to improve search engine visibility. Journalists retain full control by reviewing and selecting the most effective options. This ensures content is both engaging for readers and discoverable online.Summaries and Meta Descriptions
Generative AI can produce concise summaries or metadata for articles, ensuring that content is accurately represented in search engines and social previews. Journalists can fine-tune these drafts for accuracy and style.Writing Alt Text
Alt text serves multiple purposes: it improves web accessibility by providing descriptive text for visually impaired users who rely on screen readers, enhances search engine optimization (SEO) by associating keywords with images, and helps contextualize images in cases where they fail to load. AI tools can draft alt text by analyzing the content of an image and generating a description.
Social Media Adaptations
AI tools can craft platform-specific versions of an article, adjusting tone, length, and focus for Instagram, Facebook, or LinkedIn. These suggestions save time and allow journalists to maintain a consistent voice across channels.Content Localization
Machine translation tools like DeepL or Google Translate can provide a first draft of translations, which can then be refined by a human editor. This approach speeds up the process while ensuring cultural and linguistic nuances are respected.Repurposing Evergreen Content
AI can identify older articles with ongoing relevance and suggest ways to refresh or adapt them for current contexts. Journalists oversee the updates, ensuring they remain timely and accurate.Drafting Standard Reports
For routine stories—such as weather, sports scores, or stock updates—AI can draft initial versions based on structured data. Journalists can then refine and contextualize the content as needed.
While the output often requires human oversight to ensure accuracy and relevance, this process significantly reduces the time spent on manually writing descriptions for each image, particularly in content-heavy workflows.
🛠️ Best gen AI tools for voices
⚠️ Note: this list of tools by AI Muse is not an invitation to purchase. Check the pricing and always read the terms and conditions.
ElevenLabs – Best for highly accurate multilingual voice clones.
Key Features: ElevenLabs is renowned for producing highly realistic and expressive voices. It offers two types of voice cloning: instant (requiring only a minute of audio) and professional (requiring more data for near-perfect replication).
Why it stands out: High-quality output, supports multiple languages, customizable voice characteristics (age, gender, accent).
Pricing: Free plan available with limited features; paid plans start at $5/month
Veed.io – Best for free AI voice cloning.
Key Features: Veed.io is primarily a video editing platform but offers a powerful free voice cloning feature. It allows users to generate up to 15 minutes of cloned audio per month.
Pros: Free option available, integrates with video editing tools.
Cons: Requires recording directly through the platform
Speechify – Best for ease of use.
Key Features: Speechify is known for its text-to-speech capabilities and also offers voice cloning. The free version allows limited cloning, while the paid version includes advanced features like audio editing.
Pros: Simple to use, high-quality output.
Cons: Limited free version; paid plan costs $99/month
PlayHT – Multilingual support
Key Features: PlayHT offers fast and high-quality voice cloning with support for multiple languages. It can clone voices using as little as 30 seconds of audio.
Pros: Fast cloning, multi-language support, API integration.
Cons: Requires longer audio samples for high-fidelity clones
Murf.ai – Best for content creators
Key Features: Murf.ai provides a comprehensive platform for voice generation and editing, making it ideal for creating podcasts or videos.
Pros: Versatile editing tools, good quality voices.
Cons: Voice quality may not match ElevenLabs