The Legal edition
The AI news of that week, in plain language.
New AI models, deepfake fraud, and a delayed AI law: lots of movement, but your final check does not change.
Deepfakes mean a document is no longer proof
Experts expect a clear rise in fraud with fake videos and fake voices this year. For your files, that changes what 'proof' really means.
Criminals increasingly use hyper-realistic AI videos and voices to misuse someone's identity. Think of CEO fraud, where a manager is imitated to approve a payment. Banks and insurers that use video or voice verification are extra vulnerable. The lesson for you: a document, a photo, or even a video call is no longer proof by itself that something really happened.
Background for subscribersAgree on one rule with your office: a document that counts as proof, you do not believe on the image alone. Ask for the original, check the source or the metadata, and verify through a second channel. This applies extra to a payment request, a changed account number, or a signature. A short video call as verification is no longer reliable enough.
AI law: strictest duties for recruitment and care delayed to 2027-2028
The stricter obligations of the EU AI Act for high-risk AI systems have been delayed. Think of AI for CV selection, exam marking, and access to care. Those duties will now only apply from December 2027 or August 2028. The transparency duties of article 50 do already apply from 2 August 2026.
Background for subscribersThis is a taster.
Subscribers read the whole Legal edition: every story, each with its background in plain language.
Subscribe →