Brand impersonation is hard to stop: AI makes attacks regenerate like Hydra

a16z Newsletter · rss · 2026-08-26

Brand impersonation is notoriously hard to detect and dismantle because information isn't "lost" and attacks occur off the brand's own platforms. A report by Doppel reveals the scale and challenges of this threat:

Status & Data

Low Discovery: Brands typically spot less than 10% of impersonations, with 94% catching fewer than half of their impersonators.
Multi-platform Coordination: Over 80% of brand impersonations occur on at least two channels simultaneously (social media, lookalike domains, mobile apps, ads, etc.). Multi-surface campaigns have grown roughly sevenfold in two years.

Takedown Difficulty

Rapid Domain Regeneration: About 60% of taken-down malicious domains are pointed at a new scam within 24 hours, and 90% within 90 days.
Sleeper Domains: Over one-fifth of malicious domains are "aged" domains that sat dormant for years before activation.
Removal Latency: While 97% of requests eventually succeed, platform response times vary, leading to a long tail of persistent threats.

Conclusion

Traditional one-off takedowns fail to solve the problem. Attackers use AI to rebuild attack vectors cheaply and at scale. Defenders need continuous, cross-surface automated monitoring.

Original post →

Brand impersonation is hard to stop: AI makes attacks regenerate like Hydra

More from Safety