NextFin

Google’s Gemini Scam Problem Shows AI Security Is Becoming a Two-Way Arms Race

Summarized by NextFin AI
  • Scammers are leveraging Gemini AI to create spam messages, impacting Google's ad business which has already blocked over 8.3 billion ads in 2025.
  • Google's Gemini AI is shifting ad screening from static rules to model-driven judgment, increasing the complexity of policing ads.
  • Generative AI allows scammers to produce more convincing outreach, leading to a disproportionate increase in review work for Google.
  • Google aims to protect its $200 billion annual ad revenue by enhancing ad safety, but faces challenges as scam campaigns become more sophisticated.

NextFin News - Scammers are using Gemini AI to build spam messages, Google said on June 12. The harder fact for investors is that this is hitting a business that had already blocked or removed more than 8.3 billion ads in 2025, including 602 million tied to scams, while suspending 24.9 million advertiser accounts.

This is not about whether bad actors have discovered AI; they already had. It is about the cost of policing Google’s ad business as generative tools make fraud cheaper to produce and harder to spot. Google says Gemini is now embedded in its enforcement systems, analyzing account age, behavioral cues and campaign patterns rather than relying as heavily on older keyword-based filters. On the surface this looks like another AI-safety warning; the real issue is that ad screening is shifting from static rules to model-driven judgment at industrial scale.

That changes the economics on both sides. A scammer can use generative AI to produce more convincing outreach, test variations quickly and target specific users without building a large human operation. For Google, even a modest increase in scam quality can create a disproportionate increase in review work across billions of ads and millions of websites. The real trade-off is not growth versus safety in the abstract; it is precision versus friction. Catch too little and users lose trust. Catch too much and legitimate advertisers absorb the cost through delays, false positives and account suspensions.

Google’s own figures show why the company is leaning so hard on AI. Keerat Sharma, Google’s vice president and general manager of ads privacy and safety, said in April ad-safety report coverage that Gemini helps detect and block harmful ads in real time, while Google’s models analyze hundreds of billions of signals. Google also said Gemini helped cut incorrect advertiser suspensions by 80% and let teams act on more than four times as many user reports in 2025 as the year before. Those are not cosmetic improvements. They suggest the company is trying to protect the margin structure of a digital ad machine with more than $200 billion in annual ad revenue reported by industry trackers, where trust-and-safety failures can either scare off advertisers or force heavier manual review. Whether that works depends on whether those gains hold as scam campaigns become more polished, personalized and adaptive.

The beneficiaries are clear enough: large platforms with enough compute, data and engineering depth to run an AI-against-AI defense. The pressure falls on smaller advertisers that may still get caught in stricter automated reviews, and on rival platforms including Microsoft, Meta and OpenAI, which face the same abuse pattern as consumer-facing AI spreads. The risk nobody is talking about is that safety spending may keep rising alongside adoption, not fade as models improve. Google’s disclosure does not show Gemini is uniquely exposed; it shows the abuse vector is structural. The math doesn’t add up yet if investors assume AI lifts engagement and output without also locking platforms into permanently higher enforcement costs.

Explore more exclusive insights at nextfin.ai.

Insights

What are the origins and key concepts behind Google's Gemini AI?

How does Gemini AI integrate into Google's current ad enforcement systems?

What feedback have users provided regarding the effectiveness of Gemini in detecting scams?

What recent updates have been made to Gemini AI's capabilities as of 2025?

What are the potential long-term impacts of AI-driven ad enforcement on smaller advertisers?

What challenges does Google face in balancing ad safety and legitimate advertiser needs?

How do Google's ad enforcement costs compare to those of its competitors like Microsoft and Meta?

What structural issues contribute to the ongoing abuse of AI in advertising?

What are the implications of Gemini AI reducing incorrect advertiser suspensions by 80%?

How does the shift from keyword-based filters to model-driven judgment affect ad screening?

What are the risks associated with rising safety spending in the AI advertising landscape?

What trends are emerging in the use of generative AI for creating scams?

How does the effectiveness of Gemini AI impact user trust in Google's ad platform?

What are the potential future developments in AI-driven ad enforcement technologies?

What strategies might competitors adopt to counteract the AI-generated scams?

How does Gemini AI's performance in 2025 compare to previous years?

What ethical considerations arise from AI's role in policing digital advertising?

Search
NextFinNextFin
NextFin.Al
No Noise, only Signal.
Open App