AI in Journalism Controversy: Harvard Professor’s FT Column Triggers Backlash and Review

What Actually Happened

Ricardo Hausmann, a professor at Harvard’s Kennedy School of Government and director of the Harvard Growth Lab, wrote an opinion piece for the Financial Times titled “Trump is taxing the dark matter that pays America’s way.”

After publication, the FT added an editor’s note stating: “It has come to our attention that AI was used to condense a longer draft of this column prior to submission to the FT and our own editorial involvement. The FT editorial code of conduct specifically prohibits the use of AI in the writing process.”

The FT confirmed to reporters that it would conduct an editorial review but declined to elaborate beyond the note itself. Harvard and Hausmann did not immediately respond to requests for comment.

The Online Reaction Was Swift

The piece drew immediate criticism from prominent voices across media and academia.

• Bloomberg’s Joe Weisenthal described the situation as “astonishing, both in its writing and its publication.”
• Dartmouth political scientist Brendan Nyhan shared a screenshot from Pangram, an AI detection tool, showing 71% of the text flagged as AI-generated.
• A Harvard student and freelance writer, Alex Bronzini-Vender, called it “the worst AI slop I’ve seen in a major publication.”

The reaction wasn’t just about the AI use itself — it was about the gap between the author’s credibility and the apparent quality of the output.

Why This Case Cuts Deeper Than Most

This isn’t a story about a content farm or an anonymous blog post. It involves a named academic at one of the world’s most recognized institutions, writing for one of the most respected financial publications in the world.

That context matters. Readers and editors extend trust based on bylines. When AI is used without disclosure — even in a condensing or editing capacity — that trust gets complicated fast.

The FT’s policy is explicit: AI is prohibited in the writing process. The question the review will likely examine is where “condensing a draft” sits within that definition.

The Disclosure Problem Is Getting Harder to Ignore

Most AI writing policy debates focus on full generation — did a human write this or did a machine? But this case highlights a murkier middle ground: AI-assisted editing, summarizing, or restructuring of human-written drafts.

That’s a workflow many professionals are quietly using right now. The Hausmann case makes it harder to treat that as a gray area.

For newsrooms and publications, the practical implication is clear: policies need to address the full spectrum of AI involvement, not just outright generation. “AI was only used to condense it” is likely to become a common defense — and publications need to decide in advance whether that’s acceptable.

The Broader Education Angle

The controversy lands at an awkward moment for higher education. Universities are actively debating how to handle AI use among students, with many professors tightening exam conditions and submission rules.

Harvard senior research fellow Christopher Dede has argued that institutions are focused on the wrong problem — that the goal should be preparing students to work with AI effectively, not just preventing them from using it.

That’s a reasonable long-term argument. But it doesn’t resolve the short-term tension: when a professor at that same institution uses AI in a way that violates a publisher’s code of conduct, the credibility of any institutional AI guidance takes a hit.

What This Means for AI Tools in Professional Writing

AI editing and condensing tools are genuinely useful. They can tighten long drafts, improve flow, and save time. The problem isn’t the capability — it’s the absence of clear disclosure norms around how and when they’re used.

Right now, most publications are playing catch-up. Policies written before generative AI became mainstream don’t map cleanly onto today’s workflows.

The practical takeaway: If you’re using AI at any stage of a piece destined for a publication with an editorial code, check the policy before you submit — not after. “I only used it to condense” is not a safe assumption. And if you’re running a publication, your AI policy almost certainly needs an update that addresses assisted editing, not just full generation.

The FT’s editorial review of this piece will be worth watching. How they define the violation — and what they do about it — could set a visible precedent for how major outlets handle similar cases going forward.