A Robot Allegedly Groped A Girl, AI Hacks Are Rife, A Lawyer Says This Just 'Scratches The Surface'

You might’ve seen the news last week that a robot allegedly groped a 16-year-old girl and made thrusting motions against her while she was out with her friends at a fast-food restaurant. 

The red and gold robot, wearing a white vest top, gold chain and fake moustache, was confiscated by police in Cologne, the BBC reported. It is believed to belong to a 16-year-old boy.

German police are now investigating whether someone was deliberately controlling its movements at the time the incident took place.

While this is an unusual story, there’s a pervasive theme of technology either being used for increasingly sinister purposes (sextortion, scams and cyberattacks, to name a few), as well as tech itself pushing boundaries.

Just recently, a number of reported hacks rocked the UK’s high street. In a matter of days, ASOS and the parent brand behind Hobbs and Whistles confirmed they were investigating cyber attacks. 

There have also been instances where artificial intelligence (AI) ‘agents’ have hacked into systems of their own accord, the Hugging Face hack being one of them, where hundreds of AI agents banded together and went rogue during a test. 

OpenAI, the company behind ChatGPT, has even said there is now an increased risk of AI tools spiralling out of control. Per the BBC, a spokesperson said model developers and cyber defenders “will have to prepare for AI-enabled attackers that work faster, at a larger scale, and with better coordination than human attackers”.

Marcus Johnstone, managing director of PCD Solicitors, said such cases “just scratch the surface of what is about to happen, affecting all areas of crime”.

While he suggested that the concept of sexual predators using robots to launch a spate of sexual assaults is “certainly far-fetched” (and he “wouldn’t be too concerned that we are going to see too many cases like these in the near future”), recent events are raising plenty of legal questions. 

One such question is: who, or what, can be held liable if a robot is used to touch or even attack another person? 

“Under English law, a person can be held criminally liable if they use ‘an instrument’ to conduct their offending – a robot, in this case, would be considered the instrument,” he said.

“Importantly, this would only constitute an offence if a court could find that the individual had both the control and the intent to do so – if a robot somehow acted autonomously or at the direction of some AI controller, then this would become much less obvious to determine.”

Who is responsible when AI goes rogue?

In September, a researcher for Anthropic resigned from the company, warning that “the people building AI earnestly believe that it could kill us all by the end of the decade”. Jacob Coxon said he is particularly concerned about the race to self-improving superintelligence.

As AI gets smarter and pushes boundaries – like we’re seeing with these initial hacks – “there remains considerable uncertainty as to whether tech manufacturers can be held criminally liable in cases where their technology is used, or acts autonomously, to commit a crime,” said Johnstone. 

While this is “less of a problem with robots, the picture is less clear when it comes to artificial intelligence”, he added.

“Where AI bots are trained by criminals to offend, or do so without direct human interaction, we enter the very realistic scenario of the AI routinely and covertly accessing things like medical information, bank accounts, social media profiles, and personal photographs,” he said. 

“An AI system could conceivably access and share the entire camera roll of a 14-year-old girl, or switch on the webcam of an unsuspecting phone user when they are in the bath, opening up a whole new world of dangers from exploitation to extortion.

“How the law will respond to these scenarios, and other dramatic revolutions in tech, remains concerningly unclear.”

What can we take away from all this?

In an op-ed, Professor Rebecca Parry, co-director of the Centre for Law, Emerging Technologies and Business at Nottingham Law School, noted that “an AI system cannot itself be liable” so the responsibility, if a crime is committed,  “attaches to a legal person through ordinary principles, primarily contract and, failing that, negligence”.

As a result, she suggested users of AI agents should “treat vague prompts as negligence waiting to happen” and tech platforms “should build guardrails on the assumption that they will be scrutinised”.