Security
AI-Enabled Financial Fraud Is No Longer a Future Problem
The FTC flagged a "big wave" of AI-assisted IRS scams. A Hong Kong worker lost $25M on a deepfake video call. Q2 2025 saw $347M in deepfake fraud losses in three months, up 41% quarter-on-quarter. Voice cloning now requires only a short audio sample and an afternoon. The technology that lowered this barrier has no specific federal fraud statute written against it yet.
Read full story →
Research
What We Actually Know About AI Productivity Gains — and What We Don't
Anthropic released two new research papers on how AI changes work. The 14% gain figure and the 50% figure are both real — and measure completely different things. Software developers capture the largest share of benefit. But AI assistance may also slow skill formation in junior workers. The evidence is accumulating; the certainty behind the biggest claims is not yet warranted.
Read full story →