What we do
Parameter covers artificial intelligence and the technology industry around it. Our job is narrow and specific: read across the reporting on a story, work out what is actually being claimed and by whom, and publish one clear article with the sources attached.
Sourcing varies by story, and we show you which you are reading. Some articles are corroborated across several independent outlets or carry the primary announcement itself. Others rest on a single report, because that is all that exists yet. Every article lists the sources it was built from, so you can judge the difference rather than take our word for it.
That is deliberately different from an aggregator. If ten outlets cover the same model release, an aggregator gives you ten links. We give you one story that says what was announced, what is still unclear, why it matters, and where every claim came from.
Who we are for
Engineers, founders, researchers, policy people and anyone who needs to follow AI closely without reading twenty newsletters. We assume you are technically literate and short of time.
We cover the industry globally, and we pay particular attention to where the rules are being written. The EU AI Act, US federal and state regulation, and export controls on compute now shape what gets built and who can build it at least as much as any research result. That story is usually reported as politics or as technology, rarely as both, which is the gap we intend to close.
What we cover
- AI Models — Frontier and open-weight model releases, benchmarks, context windows, pricing and capability changes.
- AI Tools — AI products people actually use — writing, design, productivity, search and workflow tools.
- AI Agents — Autonomous agents, agent frameworks, tool use, computer use and multi-agent systems.
- Open Source AI — Open-weight models, permissive licences, community fine-tunes and open AI infrastructure.
- AI Coding — Coding assistants, code generation, autonomous software engineering and developer tooling.
- AI Image & Video — Generative image, video, audio and 3D models, plus creative-industry impact.
- AI Business — Funding, enterprise adoption, revenue, partnerships, regulation and the economics of AI.
- AI Research — Papers, evaluations, interpretability, safety research and results worth understanding.
- Robotics — Humanoids, embodied AI, autonomous vehicles, drones and industrial robotics.
- AI Hardware — GPUs, accelerators, data centres, chips, memory and the compute supply chain.
- AI Apps — Consumer AI apps, mobile assistants, wearables and how people use AI day to day.
- AI Policy — Regulation, safety rules and export controls — the EU AI Act, US federal and state law, and the compute restrictions that decide who can build what.
- Technology — Broader technology news that shapes the AI industry — cloud, security, platforms and policy.
How we work
Our newsroom runs an automated pipeline for discovery, deduplication, fact extraction and checking, with AI assisting research and drafting. Quotations are checked against the source material, figures must trace to a cited source, and a draft that fails either check does not run. Corroboration across independent sources is recorded and shown to you, not required — some stories only have one source, and we say so rather than hiding it.
Nothing publishes itself. The software can research, draft and check; the highest status it can assign is “approved”, which puts a story in a queue. Moving a story from that queue onto this site is something a person does, and only a person can. We think readers should know precisely how that works, so we have written it down in full rather than reduced it to a disclaimer. Read the editorial policy, and the list of sources we monitor.
Who publishes this
Parameter is an independent publication, not owned by or funded by any of the companies it covers, and it carries no sponsored content. Editorial responsibility rests with the publication, and every enquiry reaches the person who holds it through the contact page. Stories are bylined to the Parameter editorial team; where a named writer is responsible for a piece, that name is on it. We do not invent bylines and we do not attach a person’s name to work they did not do.
When we get something wrong we correct it on the article and list it on the corrections page, with the date and what changed. We do not quietly edit errors away.
What we are not
We are not doing original on-the-ground reporting, and we do not present other newsrooms’ scoops as our own. When a story exists because a reporter somewhere did the work, we say so and link to them — and we never write “we tested” or “our benchmark” about work that was somebody else’s. We are also not a place for AI hype: a demonstration is a demonstration, a benchmark is a benchmark, and we try to say which is which.
Get in touch
Story tips, corrections, press releases and feedback all go through the contact page. Corrections are published on our corrections page.