‘Tailored rules’ for AI chatbots ‘would add to Online Safety Act complexity’

Lauro Fava of Pinsent Masons was commenting after the UK government confirmed its intention to table an amendment to the Crime and Policing Bill “to allow the government to require chatbots not currently in scope of the Online Safety Act to protect their users from illegal content”.

The Online Safety Act is wide-ranging and detailed legislation that, among other things, requires certain online service providers to remove illegal content from their platforms. For certain types of “priority” illegal content and content that is harmful to children, the providers have an obligation to proactively monitor their platforms and remove this content before users encounter it. The greatest obligations fall on providers of certain high risk and high reach services, under a categorisation system for regulated services. The government has estimated that at least 100,000 online services, based in the UK and overseas, are in-scope of the Act.

The Crime and Policing Bill is in the latter stages of the parliamentary process. It completed its journey through the House of Commons last summer and has subsequently reached the report stage in the House of Lords.

The plans to act in relation to AI chatbots reflect a broader government intention to act iteratively to address online safety issues. It said it intends to use new legal powers so as it can “act fast … within months”, where the evidence points to a need for action.

In this regard, the government has suggested further potential legislative interventions could arise from a consultation it will open next month on children’s wellbeing online. That consultation will “confront the full range of risks children face online”, it said.

Among the specific interventions that the government has said it will consult on are restrictions on “infinite scrolling” and the introduction of a minimum age limit for social media. The consultation will further explore what technology companies can do to stop “nude images of children” being shared by or with children, while the government is also considering restrictions on children’s use of AI chatbots and options to age restrict or limit children’s use of virtual private networks (VPNs). Online platforms could also face new restrictions on the deletion of children’s data in the event of their death, according to the government’s statement.

Fava said: “The complexity of the Online Safety Act means that it may not be straightforward to simply add AI chatbots to its scope. Its rules are designed specifically for social media and search services. It is likely that a tailored set of rules would need to be created.”

“There is undoubtedly a need for the legislative process to move faster. The Online Safety Act took many years to go from inception to enactment. But new laws should still be based on thorough research and consultation. This is what ensures that laws achieve the intended objective without having unplanned consequences. The government is acknowledging this,” he said.

“The government should also look for ways to streamline new requirements. The Online Safety Act is a very complex piece of legislation, yet it is still being criticised for being inadequate. Laws that focus on the objectives platforms are expected to achieve, rather than prescribing complex and detailed rules, are much less prone to being outpaced by technological advancements. They can be quicker to enact, and they would enable platforms flexibility to identify solutions which work better for them,” Fava added.

Fava said that if the government took forward plans to regulate VPNs, it would likely attract controversy.

“These are services designed to protect their user's privacy online. Requiring them to verify the age of their users could undermine the purpose of the service,” he said.