Google just shut the front door on its own open source bug bounty program. The reason was not budget or legal risk. It was volume — a flood of AI-generated vulnerability reports that looked credible and were not. If that failure mode sounds familiar, it should: it is the same thing happening to your contact forms, your pitch inbox, and your review profile right now.
“This pause is due to a significant rise in automated submissions, the vast majority of which are not valid.”
Google, quoted by Anthony Ha, TechCrunch
Our take
Strip away the security context and this is a story about intake economics. Google built a program that paid between $100 and $31,337 per valid finding, and distributed more than $81.6 million in vulnerability rewards since 2010 — including a record $17.1 million to over 700 researchers in 2025, up roughly 40% from $12 million the year before, according to BleepingComputer’s reporting. That is a mature, well-funded, heavily staffed review pipeline. It still broke. Google says it will reformat the program and report back in Q1 2027.
It broke because the cost curve inverted. For a decade, submitting a plausible bug report required skill, so volume was naturally throttled by competence. Generative tools removed the throttle. Producing something that reads like a vulnerability report now costs near zero, while verifying one still costs a trained engineer real time. When the cost to generate collapses and the cost to evaluate holds steady, any queue with a human at the end of it eventually fails. Google’s did.
Every business running a public intake channel is on the same curve. Contact forms, quote requests, job applications, guest post pitches, support tickets, affiliate signups, Google Business Profile messages — each one is a queue where generation is free and evaluation is not. Most owners have already felt it and misdiagnosed it as ordinary spam. It is not ordinary spam. Spam filters work on signatures and obvious junk. These submissions are specific, well-written, and reference your actual business. They pass every filter you own, and they waste the most expensive hour in the company: the one where somebody decides whether a lead is real.
The second-order damage is worse than the wasted hour, because it corrupts your decisions. If 40% of form fills are synthetic, your cost per lead is understated and your lead-to-close rate is understated, which means every channel decision you make from that dashboard is wrong. You will cut a campaign that was working and scale one that was feeding you noise. Agencies optimizing to form-fill volume rather than qualified pipeline are especially exposed here — the metric keeps improving while the business gets worse.
Note what Google did not do. It did not add a CAPTCHA and hope. It paused the channel entirely and committed to redesigning the intake mechanism before reopening. That is the correct instinct and it is the one most businesses will resist, because a paused form feels like lost revenue. But a form that delivers unqualified volume is not producing revenue; it is producing work. Google also kept its narrower, higher-trust paths open — patch submissions and Google Cloud repository reports still function. The lesson is not “close the door.” It is “stop accepting anonymous, zero-friction, high-cost-to-verify submissions, and keep the paths where the submitter has something to lose.”
There is a content-side read too. Google is an organization publicly declaring that it cannot distinguish machine-generated submissions from human ones at scale, in a domain with far more signal than prose. Anyone assuming Search will reliably separate thin AI content from genuine expertise should sit with that. The durable defense has never been formatting tricks — it is the stuff a model cannot fabricate: your own pricing, your own job data, your own photos, your own numbers from your own accounts.
What this means for your business
- Audit one month of form fills by hand. Pull every submission from the last 30 days and tag each one contacted / unreachable / obviously synthetic. You need the real ratio before you touch anything else. Most owners guess 5% and find 20–40%.
- Add a cost to submitting, not a puzzle to solving. Require a phone number and verify it with an SMS code, or require the visitor to pick a real appointment slot. Bots clear CAPTCHAs; they do not answer phones or show up to calendar holds.
- Stop reporting raw form fills as a KPI. Replace “leads” on every client dashboard with “contacted and qualified.” If your reporting cannot produce that number, that is the first thing to fix this quarter — not rankings, not impressions.
- Kill your open guest-post and “content partnership” inbox. It is now pure overhead and a link-liability vector. Route it to a page stating you do not accept unsolicited contributions, and spend the recovered time on your own pages.
- Re-baseline your ad accounts after cleanup. Once synthetic conversions stop firing, your cost per conversion will jump and your campaigns will look worse. That is the true number. Reset target CPA against it rather than panicking and cutting budget.
Google’s full statement and the program timeline are worth reading in the original: Google froze its open source bug bounty program due to a ‘significant rise’ in AI submissions, reported by Anthony Ha for TechCrunch.
Should you make your contact form harder to submit?
MCNM’s take: We manage Google Ads for clients, and ad platforms learn from the conversions you feed them, so junk form fills can teach a campaign to chase the wrong people. We also care about clean lead data because we once found zero tracked leads in our own data and turned the fix into a free 8-lesson course. Our view is that a smaller number of real leads beats a big number you cannot trust.
Pros
- Fewer fake leads means your team spends less time chasing people who do not exist.
- Your cost per lead and close rate start to reflect real people, so budget decisions get more reliable.
- People who finish an extra step, like picking a call time, tend to be more serious about buying.
Cons
- Some real customers will give up on a longer form, so your total lead count will likely drop.
- Reports look worse at first. Cost per lead rises once junk stops counting, which can spook owners into cutting budget.
- SMS verification and booking tools take setup time and can add a monthly fee.
Our recommendation: If junk makes up a big share of your form fills, or your ads optimize toward form fills, add friction now. If your form brings in a few real leads a month and little junk, keep it simple and keep watching. This week, call back last week’s form leads and count how many were real people with a real need.
Your turn: How many of last month’s contact form submissions would you say came from real people who might become customers? Tell us in the comments below.
Put AI to Work for Your Business
MCNM Marketing — the team behind this publication — helps businesses across Las Vegas, Southern California, and Northern Arizona turn stories like this into revenue with AI marketing solutions, SEO & digital strategy, and marketing automation. A+ BBB rated.
Discover more from Digital Media Marketing, and Technology News
Subscribe to get the latest posts sent to your email.

