European Parliament is pushing to ban several kinds of AI systems that could have a negative impact on the public. The organization expressed particular concerns about private facial recognition databases, and about the public surveillance systems used by law enforcement.
With that in mind, Parliament has passed a new resolution that calls for strong legal safeguards for the use of AI systems. Most notably, the resolution states that decision-making should never be fully entrusted to an AI system, and that a human operator should always be responsible for making the final choice. The resolution then goes on to argue that people should have access to some kind of restitution if they have been wronged.
According to Parliament, such measures are necessary to prevent discrimination, and to minimize the impact of biased algorithms. That’s especially true in law enforcement and border operations, where biased facial recognition algorithms can lead to false arrests and other adverse outcomes for people in marginalized communities. Parliament also advocates for a basic right to privacy, and opposes mass surveillance systems that track people’s movements.
In that regard, the resolution would severely limit the scope of the surveillance state. Parliament is hoping to bar any company looking to create its own private facial recognition database, and to end the police use of automated facial recognition systems in public spaces. However, Parliament acknowledged that facial recognition can help with certain kinds of investigations, and would still allow the police to use it to monitor a criminal suspect.
The private ban would apply to companies like Clearview AI, which has a database of more than 3 billion images. The police, meanwhile, would not be allowed to use any profiling algorithms that try to predict criminal behavior based on past observations, or to deploy any social scoring systems that try to assign trustworthiness ratings to people with different types of personalities.
Parliament passed the resolution with a 377-member majority, with 248 members opposed, and 62 abstaining from the vote. The legislative body also came out against the use of remote identity verification tools like biometric e-gates at international borders, and encouraged the European Commission to take legal action against states that do not adhere to AI regulations.
The European Commission has been considering a facial recognition ban for the past several years, and published a draft of a new AI bill in April. Parliament has already stated that that draft is not strict enough, and has now doubled down on that stance with this latest resolution. The European Data Protection Board and the European Data Protection Supervisor have similarly called for stronger facial recognition regulations.
October 12, 2021 – by Eric Weiss