Google Just Hired a Detective Squad
Google has deployed a new system called SAFE, short for Scaled Abuse Forensics Examiner. It was built to find AI-generated spam at a scale no human review team could match. It's the second system of its kind Google has disclosed this year, and it may be part of the September spam update.
The headlines will say "Google is cracking down on AI content." That's the wrong takeaway, and firms that believe it will make the wrong decisions.
SAFE isn't a simple AI detector that scans a page and flags robotic phrasing. It works like a forensic investigation team. And what it's investigating isn't whether you used AI. It's how you used it.
SAFE doesn't punish firms for using AI. It punishes firms for faking what AI can't give them: expertise, authenticity, and a real human presence.
How SAFE Actually Works
Google's research paper describes a team of specialized AI agents, each with a job:
The Content Agent reads for meaning and context. It looks for material that breaks the spirit of Google's policies, even when it doesn't trip any existing rule.
The Behavior Agent looks for activity that doesn't look human, like burst publishing and synchronized uploads.
The Cluster Agent maps relationships. It examines shared infrastructure to find entire networks of connected sites, not just one bad page.
The Root Agent is the orchestrator. It collects the evidence from the others and makes the final call.
That last point matters most. For years, spam systems matched patterns. If you avoided the known patterns, you slipped through. SAFE is built to catch content that dodges the rules while still violating their intent.
The loophole era is over.
Why This Should Get Every Law Firm's Attention
Legal is one of the most competitive verticals in search. It's also one of the most tempting places to cut corners. We've all seen it:
- Hundreds of "car accident lawyer in [city]" pages with the same copy and a swapped-out town name
- Blogs that publish 40 AI-written posts in a week, then go silent for three months
- Networks of thin microsites pointing back to a main firm site
- Agencies running the same templated content across dozens of client firms
Every one of those is exactly what SAFE was designed to find. Burst publishing is a behavior signal. Templated location pages are a content signal. Shared templates across a portfolio of client sites, on shared infrastructure, is a cluster signal.
That last one should keep some agencies up at night. When the Cluster Agent maps connections, it doesn't care that your firm didn't know its content was recycled. You still carry the risk.
"AI didn't make spam possible. It made spam cheap. Google just made it expensive again."
Not sure where your firm stands? Big Voodoo can review your site the way Google's new systems will, and show you exactly what to strengthen before it becomes a problem. Let's talk.
