Microsoft’s Secret AI Publishing Deal Sparks Outrage Among Academics

Key Takeaways

• Microsoft has signed a deal that lets it use content from Taylor & Francis’ journal to train AI.
• Academics have expressed concern that they weren’t told about the agreement.
• The controversy has added to rising calls for greater independence from large publishers.

An agreement allowing Microsoft to train its AI models on Taylor & Francis’s extensive catalog of scholarly publications has garnered severe criticism from academics, who say they were kept in the dark about the deal.

While publishers have increasingly turned to such data access agreements as an additional source of revenue, researchers who contribute articles to journals without payment have little to gain.

Details of Microsoft-Taylor & Francis Deal Emerge

The first public acknowledgment of the deal with Microsoft came in a market update from Taylor & Francis’ parent company Informa.

The agreement, which runs from 2024 to 2027, gives Microsoft non-exclusive access to Taylor & Francis content. In addition to payments reportedly worth £8 million annually, Microsoft will help Informa “explore AI expert applications” and “extend the use of AI within our business,” the update said.  

Taking to X to express her alarm at discovering that articles she contributed to Taylor and Francis were being published for free, Dr Ruth Alison Clemens said the deal “sets a worrying precedent for data capitalism.”

Meanwhile, the Society of Authors, a UK-based trade union for the publishing industry, said it is “concerned to see publishers signing deals with tech companies without consulting authors and creators first.”

Criticism of the agreement has added to growing calls for more independence in academic publishing, which is currently concentrated among the “big five” commercial publishers—Elsevier, Wiley, Taylor & Francis, Springer Nature, and SAGE.

Academic’s Call For Publishing Independence 

In academic publishing, debates around intellectual property rights are framed by simmering resentment for the extractive model that currently predominates. 

“Not only do these publishers not pay us for our work; they then sell access to these journals to the very same universities and institutions that fund the research and editorial labor in the first place,” lamented the McGill University Philosopher Arash Abizadeh in a recent Op-ed . 

https://twitter.com/benwurgaft/status/1814284659393511700

Abizadeh and his allies argue that the best way for scholars to break free from the Big 5 is for libraries and universities to fund journals directly, cutting out the middleman entirely. 

Independent journals owned by their contributors would also curtail large publishers’ ability to sell content to AI developers, empowering authors with greater agency to control how their work is used. 

But to reimagine the academic publishing paradigm, new platforms and distribution formats are needed.

Decentralized Publishing

In recent years, an array of decentralized platforms have emerged as alternatives to centralized publishers.

Farcaster and Lens Protocol have been proposed as decentralized answers to Facebook and X. Meanwhile, a platform like LivePeer or DTube could one day be the YouTube of Web3.

In an interview with CCN, Ciaran Murray, who is building Olas, a decentralized platform for news journalism and academic publishing, explained that these forms of media necessitate a more sophisticated approach.

While it borrows a tip-based tokenomic model from platforms like Farcaster, Olas will also incorporate a review process for fact-checking and quality control.

“The best way to do [decentralized peer review] is via a market,” Murray argues. 

Instead of automatically receiving payouts, writers using Olas Protocol must stake the money they raise, while fact-checkers must also put money on the line if they want to rate a publication. The model is designed to incentivize quality and truthfulness by crowd-sourcing evaluation and judgment in a way that discourages artificial manipulation.

Of course, Murray isn’t the only one who believes the secret to decentralizing publishing lies in tokenomics.

In a similar vein, PRINCIPIA is a hypothetical framework for peer review where authors can bid for reviewers to review their research and reviewers can offer their services at a fee. 

An even more ambitious project is PubChain, which combines PRINCIPIA’s blockchain-based bidding system with decentralized content hosting on IPFS.