The week's developments on AI, safety and ethics, explained | 05.07.24

If you’re enjoying this briefing, sign up here — and share it with a friend. Reading time: ~5 minutes

This week, tech giants hit speed bumps in Europe as strict EU rules delay their AI rollouts, Meta struggles with accurately labelling AI-generated content and concerns grow over Big Tech’s stealthy changes on their T&Cs. 

We cover this, plus how cities are using robot cats, dogs and birds to fight the loneliness epidemic and Zuckerberg’s hot take on closed-sourced AI models.

- Charlie and the Research and Intelligence Team

P.S You are now able to sign up for the super early bird offer (75% off) to the CogX Leadership Summit in London on October 7th here.

Ethics & Governance

⚖️ EU's tough stance on tech regulation: Apple and Meta have postponed launching AI products due to regulatory uncertainties. However, critics argue that overly stringent regulations could hinder innovation and leave European consumers behind in accessing cutting-edge technologies. Is the EU's tough stance on AI limiting progress?

🤖 Meta is labelling real photos as ‘Made with AI’: Photographers are discovering that even slightly edited real photos are getting flagged as "Made with AI." This confusion has forced Meta to ditch the label altogether. Now, when they suspect a post might involve AI tools, they'll use "AI Info" instead.

📱 Are Tech giants getting ‘sneaky’ with your data? They're rewriting their terms and conditions (T&Cs) to include phrases like "artificial intelligence" and "machine learning," paving the way to train their AI models on a goldmine: your public data.

🔐 Anthropic wants to lead the AI safety movement: They're launching a program to fund entirely new ways to measure AI's capabilities, with a special focus on the potential dangers. Current benchmarks just don't cut it, they say.

AI Dilemmas

🐕 Robot cats, dogs and birds to fight loneliness: The New York state is distributing thousands of robotic cats, dogs, and birds to older adults experiencing social isolation.

🚨 Children's faces in AI training data: Human Rights Watch (HRW) discovered hundreds of Australian children's photos in a massive dataset used to train popular AI image generators. These photos, scraped from the internet without consent, raise privacy concerns and the potential for misuse.

🚀Enjoying our content? We’re also on LinkedIn — Follow us to stay in the loop with the week's most impactful AI news and research!

Insights & Research

🚩 Zuckerberg had something to say about closed AI models: In a recent interview, he championed open source as the key to democratising AI tools, putting them "in many people's hands”. "I find it a pretty big turnoff," Zuckerberg said, "when people in the tech industry kind of talk about building this one true AI".

🤖 Political deep fakes top list of malicious AI use, according to a new study by Google's DeepMind. This research, the first of its kind by DeepMind, analysed real-world incidents and found deep fakes, like those recently targeting UK Prime Minister Rishi Sunak, are nearly twice as common as the next most prevalent misuse of AI.

🚀 Stop the Gen AI hype train: Rodney Brooks, an MIT robotics legend, thinks we're overhyping generative AI chatbots like ChatGPT. Brooks argues that people often see an AI perform one task and think it can do everything similarly – but that’s not the case. AI excels at specific tasks, not universal problem-solving.

🧠 ‘Expanding intelligence a millionfold by 2045’: Ray Kurzweil, a futurist and director of engineering at Google, isn't known for shying away from audacious predictions. In a recent interview with The Guardian, Kurzweil dives into his vision for the future, a future dominated by AI.

In case you missed it

Is superintelligence a threat or overhyped? Experts can't seem to agree.

✍️ Enjoying this newsletter?  Subscribe to our free weekly briefings on Preparing for AI, Cinema & AI, The Race to Net Zero, AI & DeepTech and The Future of Work

🚀 You are now able to sign up for the super early bird offer (75% off) to the CogX Leadership Summit in London on October 7th here.