The unbearable lightness of getting scammed: Future of full scale cybersecurity (www.techradar.com)

Cybercriminals are moving beyond malware and exploiting AI to scale highly personalized social-engineering attacks that prey on emotions like love, fear and loneliness. The article cites CrowdStrike’s finding that 79% of intrusion detections are now malware-free, and points to sharp rises in vishing and voice‑cloning scams (vishing up 442% last year), plus major financial impact: the FBI reports Americans lost $16B to cybercrime in 2024 (a 33% increase year-over-year), and the FTC says losses for people 60+ who lost more than $100K rose to $445M in 2024 from $55M in 2020. Organized, cross-border “fraud centers” and targeted attacks on vulnerable populations are turning scams into a large, industrialized threat vector amplified by generative AI and voice synthesis. For the AI/ML community this signals a strategic shift: defenses must go beyond classic malware detection to focus on identity protection, provenance and abuse mitigation. Technical implications include stronger authentication and liveness detection, widespread identity‑monitoring and anomaly detection, model-level mitigations (watermarking, usage constraints, better detection of synthetic media), and investment in forensic and recovery tools—mirroring home security’s layers of locks, alarms and insurance. The market is already responding: cyber insurance is projected to grow 15–25% annually through 2029. Researchers, platform builders and policymakers must prioritize model safety, detection of synthetic identities, and cross-sector collaboration to blunt AI-enabled social engineering at scale.