Tech-news BS detector: real breakthrough or hype?
Answer eight questions before you forward a headline that sounds too clean.
This tool does not judge truth or quality for you. There is no AI behind it. It simply walks through signals that make a technology story easier to trust or easier to question.
How to use the result
A green result means the story has the basic shape of evidence. It does not mean the claim is true. A warning result means you should find the primary source before you share or act on it.
A five-step verification workflow
The questions above surface signals. When you want to actually settle whether a story holds up, run it through a short workflow instead of arguing about the headline. Five steps cover most cases.
1. Find the primary source. Trace the claim back to where it began: a paper, an official release, a filing, a dataset. If you cannot reach the original thing and only find articles pointing at each other, you have not verified anything yet.
2. Separate the demo or result from the claim. A story often shows one narrow thing, a benchmark score, a single lab run, a staged demo, and then implies something much larger. Ask what was actually measured, and whether the sweeping claim follows from that specific result or reaches well past it.
3. Check who benefits and who paid for it. Note who is making the announcement and what they gain if you believe it. A company promoting its own product, or a study funded by a party with a stake in the outcome, is not automatically wrong, but it warrants a closer read and a look for outside confirmation.
4. Look for independent replication or review. One group's result is a starting point, not a verdict. See whether other credible teams have reproduced it, or whether it went through peer review. A finding that only ever comes from its originators is still an open question.
5. Check the timeline. Distinguish a lab result from a shipping product. Ask when this would realistically reach ordinary use, in years if that is the honest answer. Much of what reads as breakthrough is real early work quietly dressed as something you can have today.
What a primary source looks like
Most confusion comes from mistaking a description of the work for the work itself. It helps to know the difference on sight.
A primary source is the original record. A peer-reviewed paper carries the methods, the data, and the limits the authors were willing to state. An official agency report or a company's own technical release is primary for what that body did or measured. A registered clinical trial documents what was planned and what was found. These share one trait: they are the thing, not an account of the thing.
A press release, a vendor blog post, or a secondhand news summary is a step removed. It may be accurate, but it is written to frame, promote, or simplify, and it can drop the caveats that matter most. To trace back, follow the links until they stop leading to more coverage: look for a DOI, a report number, a named institution, an author list, a date. If a claim never resolves to anything more solid than "a study found" or "the company says," you are holding a summary, and the original may say something narrower, or may not exist at all.
Common misjudgments
Even careful readers slip in two opposite directions, and knowing them helps you stay calibrated.
The first is the false negative: dismissing a real early result because it looks incremental. Genuine progress usually arrives as a modest, hedged, unglamorous finding, not a thunderclap. If you only trust claims that sound dramatic, you will wave away exactly the careful work that tends to hold up, and reward the loud stories that do not.
The second is the false positive: believing a claim because the demo is polished. Production value is cheap and proves nothing about whether a result is robust, reproducible, or ready for the real world. A slick video and a confident presenter can sit on top of a single cherry-picked run.
Staying calibrated means judging on the evidence, not the packaging. Slow down when a story feels most certain, and give a plain, well-sourced result more benefit of the doubt than a spectacular, thinly-sourced one. The tone is not the signal; the sourcing is.
FAQ
What counts as a primary source?
A primary source is the original record of the work: a peer-reviewed paper, an official report from the agency or company that did the study, a registered clinical trial, or a public dataset. A news article, a vendor blog, or a social post that describes those things is a secondhand summary, not the source. The test is simple: can you click through to the thing itself, or only to someone talking about it?
Does peer review guarantee it is true?
No. Peer review means a few qualified people checked the work before publication and did not find obvious errors. That raises the floor, but reviewers can miss things, results can fail to replicate, and some claims are later retracted. Treat peer review as a reason to take a result seriously, not as a stamp of final truth. Independent replication over time carries more weight than any single paper.
Why is the timeline question important?
Because most hype lives in the gap between a lab result and a shipping product. A demo that works once under controlled conditions can be years, or forever, away from something you can buy or use. Asking when it will realistically reach people separates a genuine early result from a headline that quietly implies it is already here.
Does this tool decide for me?
No. There is no AI behind it and it does not judge whether a claim is true. It only walks you through the same questions a careful reader asks, then points out which evidence signals are missing or worth a second look. The final call, and the work of checking a primary source, stays with you.
What if I cannot find a primary source at all?
That absence is itself a signal. A real result almost always leaves a trail: a paper, a filing, an official page, a named research team. If every link leads back to the same press release or an unattributed claim, treat the story as unverified rather than confirmed, and hold off before sharing or acting on it.