Browse AI Says Access Denied: Troubleshooting Blocked Websites
Learn what Browse AI’s “site blocked the preview” message means, how to check training, cookies, and CAPTCHA issues, and what to try next.
If Browse AI shows “The site blocked the preview,” Browse AI says the target site is refusing automated visits. That does not by itself mean your robot is broken or Browse AI is experiencing a general outage. First identify whether the task needs visible data only, interaction or login, or a supported CAPTCHA flow. Then follow the matching steps below. Some sites do not permit automated access, and the documented options do not guarantee that every block can be resolved.
1. Identify what was blocked
Read the exact preview or run error before changing the robot. Browse AI distinguishes an explicitly blocked preview from other problems, such as a page that did not load, an incorrect starting URL, data that is not present on the selected page, or an internal preview error. If the message says the site blocked the preview, use the blocked-site route; retraining without identifying the failure may not help. Browse AI’s guide to choosing a training method describes this message as the target refusing automated visits.
- Save the exact error text and the URL being previewed.
- Check whether the same URL loads normally in your own browser. This is a diagnostic comparison, not proof that automated access is allowed.
- Note whether the site requires a login, a click, typing, navigation, or a CAPTCHA before the requested data appears.
- Check whether the failure occurs during training preview or during a robot run. CAPTCHA support, for example, has different availability during creation and execution.
2. Choose the training method that fits the page
Use the least complicated Browse AI workflow that matches what the robot must do:
| What the task needs | Browse AI route |
|---|---|
| The requested data is already visible on the page | Start with Table Studio. |
| The robot must click, type, navigate, or log in with a password | Use Robot Studio. |
| The site remains blocked after Table Studio and Robot Studio cannot train it | Browse AI directs users to contact support; its guide also points to the Chrome extension. |
These are workflow recommendations, not a promise that the site will allow automation. If the page’s content is already visible, interacting with unrelated controls during training can add brittle steps. If the data is behind a login or interaction, a table selection alone may not reproduce the path to it. See Browse AI’s training guide for its training workflow details.
3. Refresh an expired login session
A robot that previously worked on account-only content may stop when its session expires. Browse AI’s failure guide recommends refreshing the session cookies in the robot settings and testing again:
- Open the robot’s Settings.
- Choose Update Session Cookies.
- Refresh the authenticated login session as prompted.
- Run a test task and confirm that the robot reaches the expected account page before checking its data extraction steps.
If the test still fails, check whether the account requires an additional verification step or whether the site has changed its security behavior. Do not treat repeated retries as a fix for a site that is deliberately restricting automated visits. Browse AI documents cookie refresh as a response to an expired session, not a way to bypass a site restriction. Browse AI’s robot failure guide covers session cookies and retraining after site changes.
4. Handle cookie notices without recording fragile steps
A cookie banner can interrupt training when it shifts position, appears only in some regions, or varies between visits. Browse AI advises leaving a cookie notice untouched during training when it does not obstruct the data. For account-based sites, its documentation describes using the existing browser session cookies. See Browse AI’s cookie pop-up guidance.
- If the notice does not cover the data, avoid adding a click on it to the recorded workflow.
- If it covers the target content, determine whether it is a one-time consent screen or a variable banner and account for that behavior in the training path.
- For a logged-in workflow, refresh the session cookies if the session has expired; do not confuse a stale login with a cookie-consent notice.
5. Check CAPTCHA and site security limitations
CAPTCHA support is limited. Browse AI says it supports some standard reCAPTCHA and hCaptcha challenges in some circumstances, but not custom CAPTCHA systems. Its dedicated CAPTCHA instructions say the “Solve CAPTCHA” capability must be enabled by Browse AI support for specified robots, is available on paid plans, and applies during execution rather than Robot Studio creation. A challenge shown while creating a robot is therefore a different case from an eligible challenge during an enabled production run. Read Browse AI’s CAPTCHA instructions and its compatibility guide.
Browse AI also says stronger bot detection can prevent access, particularly on high-security sites that require login. The vendor notes that sites may observe logins from multiple IP addresses and frequent robot tasks. Do not assume that proxies, changing IP addresses, or repeated attempts will solve the problem; the documentation does not guarantee that outcome. If the challenge is custom or the site refuses automated visits, contact the site or Browse AI support for an allowed route.
6. Retrain or escalate based on the symptom
- Page structure changed: retrain a custom robot against the current page and verify each selected data point.
- Authenticated page stopped loading: update session cookies and test the login path.
- Prebuilt robot failed: Browse AI recommends contacting support for prebuilt robot problems.
- Custom robot keeps failing after retraining or needs frequent retraining: Browse AI recommends contacting support.
- Preview explicitly blocked after Table Studio and Robot Studio fail: contact Browse AI support; its training-choice guide also recommends trying the Chrome extension in this situation.
Keep a short record of the URL, timestamp, exact error, whether login was required, whether a CAPTCHA appeared, and the training method used. That makes it easier to distinguish a stale session or changed page from a site-level automated-access restriction when asking support for help.
7. Troubleshooting by error pattern
| Symptom | Likely cause | Next step |
|---|---|---|
| “The site blocked the preview” | The target refuses automated visits, according to Browse AI. | Choose the training route based on the task; if both visual training methods fail, contact support or try the Chrome extension as Browse AI recommends. |
| Page does not load | Could be a temporary page problem, wrong URL, or access restriction; this is not the same diagnosis as an explicit blocked-preview message. | Verify the URL and the page in a normal browser, then record the exact robot error. Escalate if the target continues to restrict automated visits. |
| Robot worked before, then stopped on a login page | Session cookies may have expired. | Use Settings → Update Session Cookies, then run a test task. |
| Robot fails after a site redesign | Selectors or page structure may have changed. | Retrain the custom robot and validate the selected fields. |
| Cookie notice interrupts recorded clicks | The banner may vary by visit or location. | If it does not cover the data, train without interacting with it. For authenticated pages, check the session separately. |
| CAPTCHA appears in Robot Studio | CAPTCHA solving is not available during robot creation. | Check whether the challenge is a supported standard type and contact Browse AI support about eligibility for execution-time enablement. |
| Custom CAPTCHA or persistent security wall | The challenge may be unsupported or the site may prohibit automated access. | Do not assume retries will bypass it. Ask the site for an approved access method or contact Browse AI support. |
| Requested fields are not found | The selected page may not contain the data, or the training workflow may not match the needed interaction. | Confirm the data is visible at the chosen URL; use Robot Studio if clicks, typing, navigation, or login are required. |
8. Reliability and cost considerations
Reliable monitoring depends on keeping the robot’s access path stable: use the correct training mode, refresh expired sessions, and retrain after meaningful layout changes. A successful browser visit does not establish that a site permits automation, and Browse AI’s documented methods do not guarantee access in the face of bot checks or unsupported challenges.
For planning, count the maintenance work as well as task runs: authenticated sessions can expire, cookie notices can vary, and page redesigns can require retraining. CAPTCHA solving has eligibility and plan constraints described by Browse AI, so verify those with its support and current account details before relying on the feature. Browse AI’s compatibility article dated February 20, 2025 claims a success rate of over 90% for newly created custom robots. That is the vendor’s own claim, not an independently verified rate, and it does not measure access-denied cases specifically.
9. Screenshot a page without configuring a browser
If your immediate goal is a visual record of a page, ScreenshotNeo provides a website screenshot API and MCP server. It can return PNG, JPEG, WebP, or PDF from a GET request. This does not extract structured data or grant access to a site that blocks the request.
Or skip the browser setup
Make one request with your API key and target URL. The examples save or retrieve the response body; use an appropriate file extension for the output format you request. See the ScreenshotNeo API documentation for request options.
curl -G "https://api.screenshotneo.com/v1/shot" -d access_key=YOUR_API_KEY --data-urlencode url=https://stripe.com -o shot.webpimport requests
r = requests.get(
"https://api.screenshotneo.com/v1/shot",
params={"access_key": "YOUR_API_KEY", "url": "https://stripe.com"},
timeout=90,
)
open("shot.webp", "wb").write(r.content)const q = new URLSearchParams({ access_key: 'YOUR_API_KEY', url: 'https://stripe.com' });
const res = await fetch(`https://api.screenshotneo.com/v1/shot?${q}`);
if (!res.ok) throw new Error(`Screenshot request failed: ${res.status}`);
await Bun.write('shot.webp', res);ScreenshotNeo removes cookie and consent banners, newsletter popups, and chat widgets before capture; each step can be turned off. Bot checks/CAPTCHAs, blank pages, timeouts, and failed loads are not billed, and cache hits are not billed; responses include X-Page-Verdict and X-Billed headers. Its MCP server has take_screenshot, get_page_info, and capture_pdf tools for AI agents. The free plan includes 1,000 screenshots a month with no card; paid plans start at $5 for 3,000 shots. See ScreenshotNeo and the documentation. Sign up free for 1,000 screenshots a month, no card required.
Frequently asked questions
Does “access denied” mean Browse AI is down?
Not necessarily. Browse AI describes the specific “site blocked the preview” message as the target refusing automated visits. Check the exact error before diagnosing a broader service problem.
Can I fix every blocked website?
No. Browse AI documents unsupported CAPTCHA types and site security restrictions, and it does not promise every target can be accessed.
Does updating session cookies solve CAPTCHA?
No. Cookie refresh addresses an expired authenticated session. CAPTCHA support has separate type, plan, enablement, and execution-time limits.
Should I keep retrying or switch IP addresses?
Browse AI notes that sites may observe multiple-IP logins and frequent robot tasks, but its documentation does not present IP changes or repeated retries as a guaranteed remedy. Identify the failure and use the documented escalation path.
