US President Joe Biden wants to issue an executive order requiring AI vendors to conduct certain tests in the future if the programs pose risks to national security and public health and safety. The US presidential office announced this on Monday. “It’s the next step in an aggressive strategy to do everything we can to realize the benefits of AI and mitigate the risks,” said Deputy Chief of Staff Bruce Reed.
According to regulations, before new AI software hits the market, the results of these tests designed by authorities must be presented to the government. Additionally, the US Department of Commerce is expected to develop guidelines for labeling AI-generated content. So far, companies such as ChatGPT provider OpenAI, Alphabet subsidiary Google and Facebook parent Meta have voluntarily committed to labeling AI content. This commitment is not enough for the US government.
debate
How many rules does artificial intelligence need?
The decree should also cover topics such as data protection, civil rights, consumer protection, labor law and scientific research. The US government said it should be signed on Monday. Biden has declared that AI is one of his top topics and that it requires action as quickly as possible. The US government has acted too slowly in areas such as social media, which is now causing problems among young people. Apparently, the US is planning to introduce corresponding legislation in the US Congress.
G-7 promotes voluntariness
The G-7 also wants to present a “Code of Conduct” for AI on Monday. According to Politico, it should be a voluntary commitment, but Canada, France, Germany, Italy, Japan, Great Britain and the USA want to use their support to get companies to sign the rules of conduct, writes the online newspaper.
The rules also came about at the initiative of former EU competition commissioner Margrethe Vestager, who considered a voluntary agreement, rather than a legislative initiative, better suited to gaining the broadest possible approval. European Commission Vice President Vera Jourova sees the G-7 Code of Conduct as a temporary solution until regulations are in place, she said at a recent conference in Kyoto.
more on the subject
AI pioneer urgently warns of danger
According to Politico, companies that adhere to the G-7 rules of conduct are committed to minimizing AI risks caused by potential vulnerabilities. Potential limits must also be identified and safety checks implemented. It is also planned to create a regulatory system for data protection, among other things. Generative AI, i.e. “learning” systems like ChatGPT that generate text and images, is also explicitly included.
The EU Parliament presented the first framework
In the summer, the EU Parliament passed the “AI Law”, the world’s first law on AI regulation. It is currently in consultation with the EU Commission and European Union member states; According to reports, a deal is anything but fixed. According to plans, the law should be finalized by the end of the year. This week, leading representatives from politics, business and research will also gather in Britain for an AI summit. Here, too, it is mainly about regulatory issues.
more on the subject
EU Parliament agrees on AI framework
According to EU plans, AI systems should be divided into different risk groups. The more dangerous the application in question, the stricter the rules that may be applied to it. There should also be complete bans on, for example, real-time facial recognition in public spaces. According to the wishes of the EU Parliament, AI systems that classify people according to their social behavior or ethnic characteristics would also not be permitted.
According to deputies, applications associated with high risk also include those that aim to influence voters in electoral campaigns. Furthermore, the right of EU citizens to defend themselves against AI applications through legal action must be strengthened.
Europeans want to unite
Germany, France and Italy have announced that they intend to work more closely on AI so that Europe can better keep up with competition from the US and China. At the same time, economy ministers from the EU’s three biggest economies warned in Rome on Monday against excessive regulation of AI in the EU. German minister Robert Habeck said that, given international competition, “strategic coordination” was needed.
Ministers have consistently highlighted that Europe can keep pace internationally when it comes to AI. Habeck said: “We don’t need to hide. We have companies that are better in many areas than the US tech giants.” At the same time, he called for faster decisions at European level. “If the wait takes three and a half years, we will no longer have a chance”, said the vice-chancellor of the Greens. “Otherwise we will end up regulating a market that no longer exists.”
Opportunities and risks through AI
In principle, AI is also seen as an opportunity to simplify work processes, for example – with the risk that jobs and knowledge could also be lost as a result. For example, AI assistants will cause programming skills to be lost in the future, resulting in “de-skilling”, an expert warned at an event in Parliament earlier this summer. It has often been criticized that there is no quality control in many AI applications.
more on the subject
Experts urge politicians to take action when it comes to AI
State Secretary for Digitalization Florian Tursky (ÖVP) said last week that increased use of AI could bring Austria around seven billion euros more added value by 2035. Tursky, also a candidate to the mayoral elections in Innsbruck, also highlighted the challenges. The federal government will therefore introduce an AI labeling requirement.
“Just as nutritional information is common for food, all Austrians should know it in the future when interacting with artificial intelligence,” says Tursky. “Our goal is to use AI solutions in the spirit of a digitally responsible society and European values,” he said and calculated in a press release: “The performance of AI systems doubles approximately every three and a half months.”