Copyright, provenance, work and opportunity

The legal position is genuinely unsettled in places. The practical position is clearer than the discourse suggests.

intermediate9 minChecked 2026-09-06

What you will be able to do

Handle AI-assisted work responsibly in a professional context, and know which questions remain open.

Before you start

  • None. Particularly relevant if you write, design or commission.

What is unsettled and what is not

Unsettled: whether and how training on copyrighted material is permitted, and on what terms. This is under active litigation and policy consultation in several jurisdictions, including the UK.

Clearer in practice: you remain responsible for what you publish. If output reproduces someone's protected expression, "the model produced it" is not a defence anyone should rely on.

Also clearer: UK law has specific provisions concerning computer-generated works, and the position on purely machine-generated output differs from that on human-authored work. If authorship matters commercially, take advice rather than assumption.

Provenance as a working habit

Keep a record of what was AI-assisted and how. It costs nothing during the work and is invaluable when a client, an editor or a court asks.

Disclose where the relationship expects it. Norms differ between journalism, academia, design and software, and the safe move is to ask rather than to guess.

Work, honestly

The effect on employment is contested and the evidence is early. What is observable is that tasks are being redistributed rather than jobs simply disappearing, and that the benefit accrues unevenly — often to those with the access, the time to learn and the security to experiment.

That distribution question is not a technical one. It is about who gets training, who gets tools, and who is expected to absorb the productivity gain without sharing it.

What good looks like

  • You have a record of what was AI-assisted.
  • You have checked the disclosure expectation for your field rather than assumed it.
  • You review output for reproduced material before publishing.

What it cannot do

  • This is general information, not legal advice.
  • The legal position is under active development and varies by jurisdiction.
  • Labour-market evidence is early and contested; treat confident predictions sceptically.

When it goes wrong

Publishing output that closely reproduces an existing work.
Check distinctive phrases before publication. You carry the responsibility.
Assuming a client is comfortable with AI-assisted delivery.
Ask at the contract stage. Retrofitting the conversation goes badly.

Privacy and data handling

Client work carries confidentiality obligations that usually predate any AI policy. Those still apply.

Sources3 referenced
  1. The government's code of practice on copyright and AI

    UK Intellectual Property Office · government guidance · link checked 2026-09-06

  2. Guidance on AI and data protection

    Information Commissioner's Office · regulatory guidance · link checked 2026-09-06

  3. On the dangers of stochastic parrots: can language models be too big?

    Bender, Gebru, McMillan-Major & Shmitchell, FAccT (2021) · peer-reviewed paper · link checked 2026-09-06

Links are checked at build time. If one has moved, tell us.

Written by
BLACK LDN
Last checked
2026-09-06
Content state
editorial

Product behaviour changes. If something here no longer matches the documentation, send a correction.

Read next

Using it responsibly9 min

Privacy and sensitive information

The question is not whether these tools are safe. It is what you are permitted to put into them, and what happens to it afterwards.

beginner

Checked 2026-09-06

Using it responsibly10 min

Bias, fairness and who bears the error

Every system has an error rate. The question that matters is who the errors land on, and whether anyone measured before deployment.

intermediate

Checked 2026-09-06