“Our Systems Have Detected Unusual Traffic”: Why Google Shows It and How to Stop It

The message appears on a page under google.com/sorry/, usually with a reCAPTCHA and an IP address printed at the bottom. That address is the important part: it is the one Google has flagged, and it is often not the one you assume.

What triggers it

Google does not publish its thresholds, and they move with context. Advanced operators like site:, inurl: and intitle: in quick succession are treated as more suspicious than ordinary queries, because that is exactly what scrapers send.

Fixing it as a normal visitor

1. 1Solve the CAPTCHA once. Google sets a cookie that exempts your browser for a while. If you keep cookies blocked for Google, you will see the page again on every search.
2. 2Turn off the VPN or proxy and search again. Consumer VPN exits are shared by thousands of people and are flagged constantly.
3. 3Check the IP printed on the page against what your connection shows. If it is not your home IP, something is routing you elsewhere.
4. 4Restart your router if your provider assigns dynamic addresses. A new IP usually clears it immediately.
5. 5Disable extensions that preview links, check prices or ‘enhance’ search results, then try again.
6. 6Run a malware scan if the message appears on every device in the house.

When it happens in automation

If the traffic Google flagged is your own script or rank tracker, the page is a precise signal: the rate from one address is too high, or the requests look too uniform. Google's terms do not permit automated queries, so for anything business-critical an official or third-party SERP API is the durable route. If you collect results yourself, the mechanics are covered in how to scrape Google search results; the short version is below.

• Spread queries across many residential exits so no single address accumulates volume. Datacenter ranges are flagged almost instantly on Google.
• Keep one exit per short session, and do not carry Google cookies from one IP to another — a consent cookie that jumps between countries is a strong tell.
• Pace like a person. Seconds between queries, not milliseconds, with jitter.
• Match the locale to the exit. hl, gl and Accept-Language should agree with the country of the IP.
• Treat the /sorry page as a stop signal. Retire that exit for the session instead of retrying through it.

def is_google_block(resp) -> bool:
    return resp.status_code == 429 or "/sorry/" in resp.url

if is_google_block(r):
    rotate_identity()          # new session tag, fresh cookie jar
    time.sleep(random.uniform(20, 60))

How long does the block last?

For a lightly flagged address, solving the CAPTCHA is enough and the block lifts at once. For an address that keeps sending automated traffic, it can persist for hours or days, and every further request extends it. There is no form to request removal; the address has to stop generating the traffic, or you have to stop using that address.

Related: CAPTCHA on every page, not just Google

If sites beyond Google start challenging you at the same time, the address is on broader reputation lists. Check it with an IP fraud score lookup. Cloudflare's version of the same experience is covered in the verify-you-are-human loop.