Is Scraping Bing Legal? Bing Terms of Service Explained

I spent an afternoon reading the Microsoft Services Agreement clause by clause to answer one question for myself: is scraping Bing legal, and what does Bing’s terms of service actually say about automated access? I had assumed the answer was a clean yes or no. It is neither. There are two separate questions hiding inside that one, and most write-ups blur them together.

The first question is contractual: does the Bing terms of service prohibit automated scraping? The second is legal: can you be sued or prosecuted for it? Those have different answers, and below I walk through both with the exact wording I found and the case law that governs it. I work in data engineering and have no legal training. Treat this as a map of the sources, and get your own counsel before you rely on it.

Does the Bing terms of service prohibit automated access and scraping?

Yes. The Microsoft Services Agreement, which governs Bing and took effect September 30, 2025, prohibits scraping in two distinct places, and one of them is absolute.

The Code of Conduct section tells users: “Don’t circumvent any restrictions on access to, usage, or availability of the Services (e.g., attempting to ‘jailbreak’ an AI system or impermissible scraping).” The word doing the work there is “impermissible,” which leaves room for argument about what counts.

The AI Services section removes that room for anything touching Copilot or the AI answers. It states: “Unless explicitly permitted, you may not use web scraping, web harvesting, or web data extraction methods to extract data from the AI services.” That is a flat prohibition with no wiggle word attached. People often search for a separate “Bing terms of use” or “Bing ToS” document covering automated queries and automated crawling, but the operative contract is this single Services Agreement. Its scraping prohibition covers crawling and automation against the service regardless of which tool you point at it.

Here is how the two clauses compare:

Microsoft tightened the AI-services language during its 2023 and 2024 updates as it folded Copilot into Bing, a change The Register reported on at the time. The practical reading is simple: the moment you accept the agreement to use Bing, scraping it puts you in breach of a contract you agreed to.

Is breaking the Bing terms of service the same as breaking the law?

No. A terms-of-service breach is a contract problem between you and Microsoft. It is a separate question from whether scraping Bing is a crime or a tort, and US courts have drawn that line clearly over the last few years.

The anchor case is hiQ Labs v. LinkedIn. On April 18, 2022, the Ninth Circuit reaffirmed that scraping publicly available website data likely does not violate the Computer Fraud and Abuse Act, the main US anti-hacking statute. The court reasoned that you cannot access a resource “without authorization” when no authorization was ever required to view it in the first place.

That ruling tracked the Supreme Court’s decision in Van Buren v. United States (2021), which read the CFAA’s “exceeds authorized access” language narrowly. The Electronic Frontier Foundation summarized the combined effect: scraping a public website, on its own, is not a federal computer crime.

So the two layers separate cleanly:

The headline most people want is here: in the US, scraping public Bing results is generally not a crime, and it is still a violation of Bing’s terms of service. Both statements are true at the same time.

What can Microsoft actually do if I scrape Bing?

The realistic enforcement is contractual and technical before it is ever a courtroom. Because scraping breaches the agreement, Microsoft’s first and easiest move is to cut off access.

In my own work pulling Bing data, the friction shows up as blocks long before any legal notice would. Bing serves automated traffic from datacenter IP ranges a CAPTCHA wall or an HTTP 429, throttles the request rate, and can ban the IP outright. I documented that whole failure pattern and the setups that get past it in how to scrape Bing without getting blocked.

The ladder of responses, from most common to least, looks like this:

Microsoft does litigate scraping when the scale and commercial harm justify it. The point for a normal data project is that the wall you hit is technical. The legal escalation is reserved for operations big enough to land on Microsoft’s radar.

The official Bing Search API is gone, which changes the calculation

For years the clean answer to “how do I get Bing data without scraping” was the official Bing Search API. That door is shut. Microsoft retired the Bing Search APIs on August 11, 2025, announced the change on May 15, 2025, and decommissioned existing instances so the product is no longer available to use or sign up for.

The official replacement is Grounding with Bing Search inside Azure AI Agents. It is built to feed real-time web data into an LLM, and it lives entirely inside the Azure platform. For a team that just wants parsed Bing SERP data in JSON, that path adds an Azure project, resource groups, and a model deployment before you see a single result.

Even when the API existed, its legal terms were restrictive in ways that surprised people. The terms required your API key to be your sole means of access, prohibited copying, storing, or caching response data beyond service continuity, and limited all returned data to “internet search experiences.” Plenty of common use cases (rank tracking, market research, training data) sat outside that permitted scope even with a paid key.

So the landscape in 2026 is:

With the sanctioned raw-data door closed, scraping the public interface or using a scraper API are the routes left for teams that need structured Bing results. Both sit in the gray zone the rest of this article describes.

What makes scraping Bing higher or lower risk?

Risk is not one number. It scales with three factors: what kind of data you collect, how aggressively you collect it, and whether the pages are public. Get those wrong and a contract-only issue can become a copyright or privacy issue.

Personal data is the sharpest edge. The moment your scrape captures information about identifiable people, EU GDPR and the California Consumer Privacy Act come into play, and those carry real regulatory fines that have nothing to do with the CFAA or Microsoft’s terms. Copyright is the second edge: result snippets and the pages behind them are often someone’s protected content, and copying substantial amounts can support an infringement claim.

Here is how I rank the common scenarios:

That last row matters because the hiQ protection depends on the data being public. Once a scrape requires authentication, you are no longer accessing a resource “for which no authorization is required,” and the CFAA analysis shifts against you. The safest reading of the case law is to stay on genuinely public pages.

For practical risk reduction on public data, the same habits that keep you unblocked also keep you defensible:

• Pull only public, non-personal results, and never anything behind a Microsoft login.
• Throttle to a human-like rate and respect signals like robots.txt as a good-faith measure.
• Cache aggressively so you re-fetch the minimum, which lowers both your footprint and your cost.
• Do not redistribute scraped content wholesale, which is where copyright exposure climbs.

How I get Bing data without managing the legal and technical edges myself

When I need parsed Bing results without building proxy rotation and CAPTCHA handling from scratch, I use ChocoData’s Bing endpoint. It accepts a query and returns structured JSON, with the IP rotation and retries handled on the server side. The request is one line.

curl "https://chocodata.com/api/v1/bing/search?q=web+scraping&api_key=$CHOCO_API_KEY"

In my runs this returns the organic listings, related searches, news, images, and answer-box data parsed and ready to use, the same fields I would otherwise have to extract from raw HTML. You can scope it to specific SERP features through the dedicated Bing organic results scraper and Bing related searches scraper endpoints, or pull rich panels through the Bing knowledge graph scraper. A scraper API does not change the legal analysis above. It handles the technical blocking so you can focus on collecting only what you are comfortable collecting.

A practical disclaimer to close on: the contract layer (Bing’s terms of service) and the law layer (CFAA, copyright, privacy) are genuinely separate, and the law varies by country. I read the primary sources for this piece in June 2026, but if you are running a commercial scraping operation at scale, get a lawyer in your jurisdiction to look at your specific use case. For the technical side of doing this without constant blocks, my walkthrough on how to scrape Bing without getting blocked and the roundup of the best Bing scrapers and APIs in 2026 cover the setups I actually run. You can also start with ChocoData if you want the parsed-JSON route.