Check a website’s scraping and AI rules before you collect the data from it.

Training a model, building a dataset or scraping from the internet for clients? RightSignal reads the terms and AI policies most scrapers/crawlers cannot understand, plus content signals, TDM opt-outs and access blocks, gives you one clear decision for your use, and assists in compliance with India’s IT law and EU’s AI Act.

Start freeView products and pricing

Online data

Signals intended to be read by humans

  • Terms of use
  • AI policy
  • Licence notice

Signals intended to be read by crawlers

  • robots.txt
  • Content-Signal
  • TDM opt-out
  • Access barriers

The RightSignal pipeline

  1. Read
  2. Assess
  3. Decide

What you gain

  • See what robots.txt misses
  • Keep more of your data
  • Respond in minutes, not weeks
  • Cut your legal bill

The problem

Most websites have rules about scraping and AI. Most scrapers never read them.

  1. The rules are scattered.

    A site can put them in its terms and conditions, an AI policy, a licence notice or a file written for crawlers.

  2. Most tools read one file.

    A typical scraper checks robots.txt at most. What the site wrote for people goes unread.

  3. The rules change.

    Sites rewrite their terms without notice. When a customer, an investor or a publisher asks months later, you need to show what the site said on the day you collected.

In India and the EU, the law gives these rules weight.

Illustration

How it works

Four steps from a web address to one clear answer.

You say what you want.

Give RightSignal the page, what you will use the data for, and whether Indian law, EU law or both apply to you.

  • Pageexample.in/news
  • UseAI training
  • LawIndia

RightSignal reads what the site says.

Three kinds of signal, starting with the ones most crawlers cannot understand.

  • Written for peopleTerms and conditions, AI policies, copyright and licence notices.
  • Written for crawlersRobots.txt, content signals, text and data mining opt-outs, AI files.
  • Put in the wayLogins, paywalls, CAPTCHAs and bot blocks. RightSignal records them and never passes one.

You get one answer.

Every page comes back with one of three answers, and the reason in plain words.

  • CollectNothing the site says refuses your use. Sometimes with conditions, such as crediting the source.
  • HoldSomething is unclear or missing. A person should look.
  • Do not collectThe site refuses your use, or a barrier stands in the way.

You keep a receipt.

A short record of what was decided, when the site was read and the exact line the answer rests on. Anyone can verify it without taking our word for it.

  • AnswerDo not collect
  • Read on10 Nov 2026
  • Rests on“may not be used to train”

A Collect is not the site owner's permission and is not a statement about the law.

Click the scene to zoom in

India and the EU

RightSignal is built for the two bodies of law that matter most to Indian AI teams.

If you build in India

No text and data mining exception, and no opt-out statute. Section 43 of the Information Technology Act asks whether you had the owner's permission, so RightSignal records where a site gives or withholds it.

If you serve the EU

A rightsholder can reserve its content from text and data mining, and the AI Act requires providers of general-purpose models to respect that. RightSignal reads reservations in files, headers, page tags and a site's own words.

If you plan to do both

The India and EU Bridge profile applies the stricter reading of the two at every step. Collect once, with records ready for both markets.

One site, three profiles

A site's reservation file could not be fetched.

  • IndiaCollect on conditions
  • EUHold
  • India and EU BridgeHold

Click the map to zoom in

Why choose RightSignal?

Collect your data with confidence. Prove it when asked.

For everyone who collects web data

  1. See what robots.txt misses.

    We read the terms, AI policies and opt-outs where most sites actually say no.

  2. Keep more of your data.

    Each use of data is decided separately, so a site that blocks AI training can still be fine for search.

  3. Breeze through due diligence.

    Show investors and enterprise buyers exactly how every source was checked.

  4. Respond in minutes, not weeks.

    If a source is challenged, its receipt number takes you straight to the data, so you can remove it or defend it.

  5. Proof anyone can check.

    Every receipt quotes the lines it relied on, and anyone can verify it without taking our word for it.

  6. Cut your legal bill.

    Your lawyers review the quoted lines and the tricky cases, not every site.

Invented example

Who it is for

Solutions built for your team

AI companies and model providers

Train on data you can stand behind.

  • Build your EU AI Act copyright policy and training summary from real records.
  • Check every source under Indian law, EU law or both.
  • Audit the data you already hold, including Common Crawl.

What you can tell your customers

“Our source records are ready for EU AI Act documentation.”

“Every source is assessed under Indian and EU law.”

RAG and AI answer builders

An index you can account for, without the manual work.

  • Confirm a source allows AI answers before you index it.
  • Skip reviewing sites by hand.
  • Watch, coming soon, will alert you when a source changes its rules.

What you can tell your customers

“We check every source before we collect it.”

Scraping agencies and data vendors

Sell data your clients can trust.

  • Ship a sealed record with every source, so clients have more confidence.
  • Win enterprise and EU buyers who now ask how sources were checked.
  • Answer procurement questionnaires with a ready source register.

What you can tell your customers

“Every item links to a sealed record of what its source allowed.”

“We never get past a login, paywall or bot challenge.”

Only if your own crawler holds to this.

AI developers

One call before every fetch.

  • Get a plain answer your code can act on.
  • Store a receipt number beside everything you collect.
  • Build and test for free before you go live.

What you can tell your customers

“We read each site’s terms, AI policy and opt-outs, not just robots.txt.”

Individuals and small teams

Clear answers, no small print.

  • One answer per page, with the reason in plain words.
  • Show clients and collaborators that you checked.

What you can tell your customers

“If a source is questioned, we can show what it said on the day we collected.”

Law firms and counsel

A new practice area, advisory and clients to be tapped.

  • Offer dataset audits, collection advice and EU AI Act copyright work.
  • Guide Indian clients into the EU, with decisions under both laws.
  • Review a target’s or supplier’s data sourcing in due diligence.
  • Work from evidence you can read, quote and cite.

What you can tell your customers

“We audit your data against each source’s own terms.”

“We review your collection plan, source by source, before it runs.”

“We build your EU AI Act copyright policy from verifiable records.”

“We help you enter the EU with your collection records in order.”

Investors and enterprise buyers

Know what you’re buying.

  • Ask for sourcing records you can verify yourself.
  • Know exactly which data to remove if a source is challenged.

Click the scene to send a sealed record across

RightSignal records what websites publish and applies it to your use. To integrate it automatically in your workflows, get access to our API or the app.

The API and Trace

One call before every fetch. Audit the data you already hold.

The API, before you collect

Add a check before the fetch, and store the receipt beside the data.

  • It is plain HTTPS and JSON, so any language can call it.
  • Store the receipt number with every item you collect.
  • The ready-made policies cover AI training under each profile and text and data mining under the EU profile.
  • You can build and test your whole integration in test mode before you run a single live check.
Request
POST /v1/gate/checks
{
  "url": "https://www.example.in/news/markets-today",
  "policy_id": "india-ai-training",
  "purpose": "ai_training",
  "crawler": "GPTBot"
}

Trace, after you collect

Find out what the sites in your dataset said, then and now.

  • You can answer a question about a Common Crawl corpus with a document and not with an assumption.
  • You receive an exclusion list that your pipeline can apply before the next training run.
  • You receive a review queue that puts the most significant rows first.
  • If you are a model provider serving the EU, you can describe in your public summary what you did about a third-party dataset and support the description with a sealed record.

Trace never needs the content of your dataset. It works from addresses and dates. Trace is available by request.

Click the scene to send a page through

Move across the data to compare then and now

The products

Four products that share one sealed record.

  • Gate, before you collect

    Know before you fetch.

    A named decision, any conditions and a sealed receipt, before your scraper takes anything.

    Early access, by invitationExplore Gate

  • Trace, after you collect

    Audit the data you already hold.

    Trace compares the evidence from collection time with what each site says today.

    Available by requestExplore Trace

  • Watch, while you rely on a site

    Know when the rules change.

    Watch will recheck the sites you depend on and keep a dated history of every version.

    Coming soonExplore Watch

  • Matters, when it counts

    Turn sealed records into an evidence pack for your counsel.

    Matters brings the relevant receipts and dated captures together for review.

    Coming soonExplore Matters

Pricing preview

Start free in test mode, and move to live seals when you need evidence to keep.

Questions people ask first

Is this legal advice?

No. RightSignal gives you evidence of what a site said and a decision under the RightSignal Algorithm. Your own lawyer advises you on what that means for your business.

Does a Collect mean the site gave permission?

No. It means RightSignal read the site's signals under your profile and found nothing that refuses or conditions your collection for that purpose. The receipt shows what was read, so you and your lawyer can judge it.

Why not ask an AI assistant whether I may scrape a site?

An assistant's answer is not evidence. It is not repeatable, it is not sealed with the wording and the time, and it may guess. RightSignal fetches the sources, quotes them word for word, checks the quotes by code and seals a receipt.

What happens when a site cannot be read?

RightSignal says so. A failed read is named as a gap, and the answer becomes a hold or a stop. It never becomes a collect.

See what your own sources say before your next crawl.