How we’re combatting AI scams with security, legislation and more

Source: Google Date: 2026-06-12 URL: https://blog.google/innovation-and-ai/technology/safety-security/combatting-ai-scams/

Summary

Google published a coordinated multi-front response to AI-enabled scams: a civil lawsuit against a China-based phishing-kit distributor operating via Telegram (the “Outsider Enterprise” network), parallel FBI action, carrier-level SMS filtering with AT&T, T-Mobile, and Verizon blocking malicious texts, Android on-device scam detection intercepting over 10 billion harmful messages per month, and advocacy for seven bipartisan US bills including the Stop SCAMS Act. The through-line is that AI is meaningfully lowering the cost and quality ceiling for fraud campaigns, and platform-level defenses need to operate at equivalent scale.

Implications

• Trust/provenance hardening: Google’s move to combine litigation, carrier coordination, and on-device detection is the first major public articulation of a defense stack that mirrors the attack surface AI has opened. The framing matters: this isn’t spam filtering, it’s provenance — distinguishing AI-generated fraud from legitimate AI-generated content at message level.
• Supply-chain: The Telegram-distributed phishing-kit angle is structurally the same problem as malicious package registries. A criminal ecosystem selling AI-augmented attack tooling as a service compresses the time between capability release and weaponization.
• Trust/provenance hardening: Legislative pressure (Stop SCAMS Act, National Strategy for Combating Scams Act) signals that platform self-regulation is losing credibility with Congress. If these bills pass, provenance and scam-detection obligations could extend to agent-generated content — which would directly affect any product shipping AI that produces outbound communications.