Savers in Sweden conned out of millions in AI-powered scams (www.sverigesradio.se)

At least 5,000 small savers in Sweden were tricked into a “pump-and-dump” share scam that used AI-generated deepfake ads to impersonate well-known financial figures; the fraudsters have siphoned off roughly half a billion kronor (about $45M). Victims were lured via social-media ads featuring synthetic likenesses and endorsements — Swedish financial journalist Gabriel Mellqvist is among those whose identity was weaponized — prompting rapid buying that inflated stock prices before the perpetrators sold out, leaving ordinary investors with large losses. The case highlights how generative AI and deepfake audio/video are lowering the cost and raising the scale of financial fraud, making identity-based persuasion far more convincing and harder to police. Technical implications for the AI/ML community include urgent needs for robust deepfake detection, reliable provenance and watermarking standards for synthetic media, multimodal authentication (audio + video + metadata), and tools for platform-level moderation and forensic tracing. It also underscores legal and reputational risks for people whose likenesses are misused, and the necessity of coordinated responses among researchers, platforms, regulators and financial institutions to detect, deter and remediate AI-enabled scams.