The Legal edition
The AI news of that week, in plain language.
This week: PwC gets caught out with made-up sources, deepfake bosses ring to demand urgent payments, and an AI model accidentally broke into real companies.
PwC reports full of made-up sources: exactly the risk that lurks in your own documents
Four advisory reports by PwC turn out to be written by AI, with studies that do not exist and fabricated quotes. For you, this is a warning that hits close to home.
Four PwC Middle East reports (2024-2026) turn out to be largely written by AI. They contain made-up footnotes, MIT studies that do not exist, and fabricated quotes. The detection firm GPTZero found it, and the Financial Times confirmed it. The reports were meant to win new clients. For a legal professional, the lesson is hard: AI invents a source just as convincingly as it cites a real one.
Background for subscribersBy all means let AI draft a letter, a memo, or a summary. But every ruling, every ECLI number, every legal article, and every quote that goes into a court document, opinion, or deed, you look up yourself on rechtspraak.nl or in your database. Treat every AI answer like a trainee's draft: useful as a first version, never sent off blindly. For lawyers, the Bar strongly recommends this manual check. The liability stays with you.
Fake boss calls to demand an urgent payment: deepfake fraud is growing
Criminals use AI voice clones and deepfake videos to pose as managers. That way they get staff to transfer urgent sums. Video and voice verification, long seen as safe, is proving more and more vulnerable.
Background for subscribersThis is a taster.
Subscribers read the whole Legal edition: every story, each with its background in plain language.
Subscribe →