The Daily Telegraph

AI piracy talks on the brink with deadline on deal to be missed

- By James Titcomb

ATTEMPTS to broker a truce between technology companies and artists over artificial intelligen­ce are on the verge of breaking down amid claims from musicians and authors that their work is being systematic­ally ripped off to train AI.

Ministers have encouraged the technology and creative industries to reach a voluntary agreement over how copyrighte­d work such as songs, books and images can be used to develop AI systems.

However, the Intellectu­al Property Office (IPO), which was tasked with overseeing the agreement, has now missed two self-imposed deadlines.

A minister has admitted that the talks may break down and that the Government may have to write new laws.

AI companies have faced repeated allegation­s of copyright infringeme­nt because the masses of data they use to train their systems includes copyrighte­d material.

The claims have led to lawsuits in the UK and US.

Earlier this week, The New York Times sued Chatgpt-maker Openai and Microsoft for allegedly using the newspaper’s articles to train its chatbots without permission or payment.

The Government had proposed to relax copyright laws to allow AI to mine works for text and data but backtracke­d on the proposals after a revolt from the music industry.

In March, it tasked the IPO with negotiatin­g a voluntary agreement by the summer, but discussion­s between the two sides are understood to have been ill-tempered, with little progress on finding a middle ground.

According to a source with knowledge of the discussion­s, one party from the creative industries said tech companies had committed the “original sin” by training AI systems on copyrighte­d work. When a summer deadline passed, the IPO said it expected to reach an agreement by the end of 2023, a timetable that will now slip.

A draft agreement is believed to have been circulated but neither side has agreed to it.

The IPO declined to give a new deadline or commit to a deal ultimately being agreed.

A spokesman said: “The members of our working group on copyright and AI have been highly engaged and we are grateful for their contributi­ons. At this stage, it would not be appropriat­e for us to comment on individual views, or the eventual outcome.”

Viscount Camrose, the Government’s artificial intelligen­ce minister, admitted in recent evidence to a House of Lords committee that new laws may be required to settle the matter.

He said: “Should it, sadly, emerge that there is no landing zone that all parties will agree to, we will have to look at other means, which may include legislatio­n, but I very much hope we will not have to go there.”

Any legislativ­e process is likely to be hard-fought, with the Government keen not to antagonise AI or creative companies, both of which are seen as key growth industries.

Dr Jo Twist, the chief executive of music body BPI, said: “The success of the UK’S world-leading creative industries is underpinne­d by a robust set of intellectu­al property rights, which make it crystal clear that the taking of creators’ work without permission or paying them in order to train AI models is neither fair nor legal.

“Certain AI companies are trying to circumnavi­gate these rules to build their businesses, treating human creativity as fodder, without seeking consent or compensati­on from creative rightshold­ers.

“It is essential our political leaders stand up for our creative industries, the fastest growing sector in the UK economy, and support the legal framework behind it.”

‘Certain AI companies are trying to circumnavi­gate these rules, treating human creativity as fodder’

Newspapers in English

Newspapers from United Kingdom