The EU’s AI Act is now in force...

Welcome back to a new edition of AI Odyssey!

Here are the top updates on everything that happened in AI this week.

1: The EU’s AI Act is now in force

It is now official; the landmark AI Act of the European Union entered into force on August 1, 2024, and is going to bring a sea change in how AI applications will be regulated in Europe. What you need to know about this risk-based regulation and its ramifications for developers and users of AI applications is contained herein.

What's the deal with the AI Act?

What the EU has done is to regulate AI applications based on their risk levels. Compliance deadlines have been backdated for a couple of years. Fast breakdown:

• Low/no-risk AI: This category includes most AI apps and will not feel the weight of regulation.

• High-risk AI: This ranges from applications like biometrics and facial recognition to education and employment-related AI. Developers would be expected to declare these systems in an EU database and meet stringent requirements in risk and quality management.

• Limited risk AI: It considers technologies like chatbots and tools that could create deepfakes. Such systems will have to meet transparency requirements so users aren't misled.

• General purpose AI: Developers of general-purpose AI systems are subject to light transparency requirements. Only the very top end models will be assessed for risks and mitigation.

Compliance deadlines…

• Six-month ban: The first deadline forbids a number of AI uses, such as the remote use of biometrics by law enforcement in public places.

• Full compliance by mid-2026: Most provisions come into full effect.

What's next for AI developers?

The finalisation of the Codes of Practice is still being worked on by the EU AI Office and is expected by April 2025. In the meantime, developers should prepare themselves for the following:

• Categorize your AI systems: Decide which of your AI systems are high-risk, of limited risk, or under GPAI.

• Understand your obligations: If you are an AI provider or deployer, understand clearly what compliance steps you need to take.

• Consult legal experts: Consultation with legal experts would be necessary to exactly understand the breadth of these regulations and ensure that every provision is fully complied with.

Industry's response: Developer of ChatGPT, OpenAI, has expressed its willingness to fully cooperate with the EU AI Office. It is readying technical documentation and user guidance on the use of its general-purpose AI models.

The Act on Artificial Intelligence of the EU will give another shape to the AI landscape, characterised by more transparency and accountability. It aspires to balance innovation with safety in the responsible development and use of AI technologies.

Presented by eMarketer

Report: Generative Search Trends 2024

Get the inside scoop on how AI is transforming search advertising in 2024. Our latest report explores Google’s AI Overviews and the rise of Meta AI. Discover why Google will command 70.5% of search ad spending by 2026 and learn how to adapt your marketing strategy for this new era. Packed with expert insights and essential data, this is your guide to the future of search.

Download the Report Now

2: Paris gives glimpse of AI's Olympic future

Artificial Intelligence is having a moment at the Paris Olympics, though it's not quite ready to take center stage. Consider this year the trailer before AI becomes a full-on Olympic participant in future games.

What's happening?

Paris 2024 is centre stage on the fields of AI. Though this will not be an event in itself, AI takes up an exciting supporting role with a list of innovative initiatives:

• Intel is an official AI partner and powers AI athlete chatbots, helping broadcasters tailor content. They're also piloting a program in Senegal to find athletic talent using AI.

• Omega: While this timekeeping giant doesn't call winners by way of AI, it nevertheless analyzes athletes' performance through technology, like swim strokes and tennis serve reactions.

• NBC An AI-powered digital Al Michaels is delivering personalized daily highlights to fans. It's a new twist on sports broadcasting.

• Airweave: Providing AI body scans in an attempt to help athletes customize their mattresses, though as always, the jury is out on bed comfort.

Big picture: The AI Olympic Agenda by the IOC charts a pathway to the future of AI in sport—improving performance and athletes' safety. The use of AI comes at a cost, which rings an alarm about how the gap between rich and poorly resourced countries will enlarge.

Not everyone's on board: While many are embracing AI, some are holding back. Warner Bros. Discovery isn't using artificial intelligence this time, preferring to wait until the technology has matured a bit more. Getty Images is being cautious as well—especially with facial recognition tech, which the business believes isn't ready for the scale of the Olympics.

The bottom line regarding AI at Paris 2024 is that it has been a glimpse of things to come. It is working backstage, yet it still hasn't had centre stage. All this will be expected in big things at future Olympics as AI matures to transform sports.

3: Argentina's AI crime prediction plan raises privacy concerns

Argentina is taking a page out of sci-fi with a controversial plan to use AI to predict crimes before they happen. But while the idea sounds futuristic, it’s also raising some serious human rights concerns.

The plan: The new security unit of President Javier Milei will analyze crime data through AI and machine-learning algorithms to predict offenses in the future. It will also work with facial recognition and monitoring on social media. Part of a tough-on-crime approach that raises controversy.

The concerns…

• Invasion of privacy: Amnesty International argues that this form of surveillance might have a chilling effect on freedom of speech because people will begin to censor themselves if they believe they are being watched.

• Abuse potential: Critics argue that it will be used against activists, journalists, and minorities, something amounting to unfair targeting reminiscent of the dark history of state repression in Argentina.

Big picture…

The Milei administration had already been under criticism for its rigid measures, with riot police being dispatched against protesters. The situation has now been compounded by the AI plan. Observers warn of militarized security policy that could erode civil liberties.

While the government insists the AI unit will be bound to respect existing laws, there is a clear potential for its misuse in this area. The balance maintained in security versus human rights would be to the fore in Argentina's adopting of AI into policing.

That’s a wrap! See you again for a fresh dose of AI Odyssey soon. 😎