AI, copyright and software

Content that uses your face or voice without consent breaks the rules of most major platforms. YouTube accepts privacy complaints about AI content that simulates an identifiable person, Google has a removal form for non-consensual fake explicit imagery, and Meta and TikTok have rules on synthetic and misleading media.

We find the content and its copies, capture dated evidence, file each report with the right form and watch for re-uploads. Where fraud, blackmail or defamation is involved, legal steps run with our partner lawyers. Acting early sharply limits how far a deepfake spreads.

It depends. Copyright generally requires human authorship: the US Copyright Office does not register material generated by AI without sufficient human control, and courts have upheld that requirement. Work that a person selects, arranges and meaningfully changes can still be protected in the parts the person contributed.

Before commercial use we check three things: the licence of the tool and plan used, documentation of human contribution, and permissions for any real people or brands that appear. We add AI clauses to agency, client and production contracts.

The EU AI Act introduces transparency rules that require certain AI-generated content, deepfakes in particular, to be clearly labelled. Brands running campaigns for European audiences should plan where labels are needed and check that their suppliers comply.

AI assistants learn about your brand from sources on the web. Behind a wrong answer there is usually an old article, an inaccurate directory or a thin Wikipedia entry. We record how different assistants answer the same questions, trace and fix the sources, make sure accurate information exists in trusted places, and measure the change monthly.