Microsoft has announced that it will back users of its AI systems in copyright claims for AI-produced work and outputs.

“As customers ask whether they can use Microsoft’s Copilot services and the output they generate without worrying about copyright claims, we are providing a straightforward answer: yes, you can, and if you are challenged on copyright grounds, we will assume responsibility for the potential legal risks involved,” an announcement by Microsoft on Thursday read.

The company’s decision appears to be the first of its kind in the AI-copyright debate, as users and companies grapple with upcoming regulations and lawsuits from creators surrounding AI outputs.

Questions concerning copyright laws have racked the world of generative AI applications, as authorits, artists, academics, and content creators have called the use of their work for training AI to be in breach of their intellectual property rights.

So far, companies have faced direct lawsuits from claimants saying that they did not consent to having their data or content used to train AI systems. Further, artists have said that AI models have copied their original style, allowing users to pass the resulting output as their own.

Microsoft’s move appears to be in the interest of its users, but also at the interest of its own company, and may be due to a lack of overriding legislation regarding AI and copyright.

In the UK for instance, has flip-flopped on AI copyright laws since the prolific rise of ChatGPT. Currently, data mining of copyrighted material is allowed in the UK for non-commercial use, such as for research purposes, but this was set to potentially expand to allow AI companies to flourish more freely in the UK. Pushback from privacy groups, however, meant that this amendment was not adopted, and the UK is still awaiting more direction for AI legislation, which may be delivered in the upcoming AI summit in Autumn. 

Further, in the US, even after a landmark case saying that AI-generated content cannot be copyrighted, this was not an outright decision on either side – the judge in this case insisted all claims be reviewed on a case-by-case basis.

Even in the EU, where the AI Act is set to go through its final round formalising it into law, copyright claims will not be straightforward when AI is concerned.

Microsoft’s move will concern users the use of Microsoft’s AI Copilots and the output they generate, specifically commercial versions of Copilot services and Bing Chat Enterprise, as well as GitHub Copilot which is used to produce code.

The announcement did come with a condition, however: users must have used the guardrails and content filters Microsoft says they have built into their products in order to qualify for legal representation.

Recommended reading

• AI, Art, and Copyright: The Great Debate Rattles On
• New Copyright Class-action Lawsuits Filed Against OpenAI and Meta
• How Could AIs Affect UK Copyright Law?

Microsoft says that these filters are designed to reduce the likelihood returning infringing content.

“These build on and complement our work to protect digital safety, security, and privacy, based on a broad range of guardrails such as classifiers, metaprompts, content filtering, and operational monitoring and abuse detection, including that which potentially infringes third-party content,” the company said.

It is unclear if there are specific filters that can filter out copyrighted material from being included in a generative AI models’ output.

However, AI companies will have to disclose what copyrighted material was used to train their systems if they are to abide by the EU AI Act.

Further, Microsoft says that users cannot attempt to generate infringing material by providing prompts or input to a Copilot service that the customer does not have the rights to use.

Microsoft did state that their new policy does not change the company’s position that it does not claim any intellectual property rights in the outputs of its Copilot services.