Wikipedia's English-language edition voted on March 20 to ban all AI-generated text from its 7.1 million articles, drawing one of the clearest lines yet between human-authored knowledge and machine-generated content.

The vote wasn't close. Volunteer editors approved the ban 44–2 after a Request for Comment process driven by mounting frustration over what editors called the "asymmetry of effort": it takes seconds to generate hallucinated, citation-free AI text and hours for human editors to fact-check and remove it. "One person can generate AI text in five seconds and post it on Wikipedia," said Ilyas Lebleu, a French AI research student who drafted the winning proposal under the username Chaotic Enby. "We can spend an hour or longer verifying everything."

The ban covers new and existing articles. Editors may still use AI tools for basic copyediting — provided the AI introduces no new content and a human reviews the changes — and for translating articles from other language editions, so long as the editor can verify accuracy in the source language. Everything else is off.

The decision formalizes what Wikipedia's volunteer editors had been fighting to enforce informally since ChatGPT's 2022 launch. The signs of AI infiltration were not subtle: articles with placeholder prompts still embedded in the text, fabricated citations, and the telltale phrase "rich cultural heritage." A WikiProject called "AI Cleanup" emerged to catalog detection techniques. Editors developed a public guide to spotting AI writing so precise that contributors can sometimes identify which model version generated a passage based on its training cutoff.

One complication: false positives. Some editors raised concerns that autistic contributors or non-native English speakers could be wrongly flagged for writing in an "AI-like" style. Lebleu addressed this in the final policy, explicitly prohibiting sanctions based on writing style alone.

German Wikipedia implemented similar restrictions in February 2026. English Wikipedia's adoption — the largest and most-linked edition — gives the policy global weight.

What Wikipedia has established is a principle other platforms have so far avoided: the source of the text matters, not just its accuracy. Whether Google, news publishers, or social platforms follow is an open question. But Wikipedia has at least made the position legible.

The post Wikipedia bans AI-generated text from its 7.1 million articles appeared first on The Media Copilot.

Content authentication standards exist. They work reasonably well in controlled conditions. The problem, according to a new Microsoft research report, is that almost nobody has deployed them at scale — and the window to fix that is closing fast.

The report, Media Integrity and Authentication: Status, Directions, and Futures, evaluates three authentication approaches — cryptographically signed provenance metadata (C2PA), imperceptible watermarking, and soft-hash fingerprinting — and finds that while the technology is mature enough, adoption remains fragmented across devices, editing tools and distribution platforms. Without broad implementation, the report warns, the gap between what AI can generate and what audiences can verify will keep widening.

The strongest finding: layering C2PA signing with imperceptible watermarking delivers “high-confidence provenance authentication” — a verifiable chain of custody from creation to publication. Fingerprinting is better suited for forensic work than real-time verification at scale.

The report introduces a concept that should concern newsrooms directly: “sociotechnical provenance attacks.” These aren’t just technical file manipulations. They exploit user perception — making real content look fake, or synthetic content look legitimate. Visible watermarks without cryptographic backing, the researchers warn, can actually make these attacks easier by training audiences to trust signals that can be forged.

There’s also a hardware problem. High-confidence authentication requires secure enclaves built into cameras and capture devices at the hardware level. Most devices don’t have this yet. Until they do, provenance claims on content captured with conventional equipment remain easier to dispute.

For media organizations, the takeaway is sobering: the C2PA ecosystem Microsoft helped co-found in 2021 now has thousands of members, but deployment in actual newsroom workflows and consumer platforms remains thin. The report frames 2026 as a critical inflection point, with regulations approaching and generative AI accelerating. Fragmented adoption now means fragmented trust later — exactly the environment where AI-driven misinformation is most effective.

The post Microsoft report: current media authentication tools aren’t ready for the AI content flood appeared first on The Media Copilot.