What caught my eye this week

TOP NEWS TO READ ??

EU Denies U.S. Pressure in AI Regulation Shift

The European Union insists its decision to scrap the AI Liability Directive was driven by a push for competitiveness, not U.S. pressure. The directive, designed to make AI companies more accountable for harm, has been dropped in favor of reducing bureaucracy. The move follows a speech by U.S. Vice President JD Vance urging Europe to embrace AI growth rather than regulation. The EU’s 2025 work program, released shortly after, confirmed the policy shift, prioritizing AI development over liability concerns. Critics question whether geopolitical influence played a role in this regulatory retreat.

Source: TechCrunch

AI Agent Hype: Boom or Bust?

Venture capitalists are warning that the AI agent startup craze may be the next tech bubble to burst, comparing it to the Web3 and Metaverse hype cycles. While investment in AI agents is soaring—European startups raised €481 million in early 2025 alone—many fear the sector is overvalued, with companies promising more than they can deliver. The challenge lies in building truly autonomous systems and proving real business impact. As Big Tech giants like OpenAI, Google, and Microsoft enter the space, startups face stiff competition. If AI agents fail to meet expectations by 2026, the bust may be inevitable.

Source: Sifted

Uber Seeks Tesla Robotaxi Partnership Instead of Competition

Uber CEO Dara Khosrowshahi wants to partner with Tesla on robotaxis rather than compete, acknowledging Elon Musk’s dominance in autonomous vehicles. With Tesla set to launch paid robotaxi rides in June, Khosrowshahi sees an opportunity for Tesla to expand its reach through Uber’s 33 million daily trips. Uber already partners with Waymo and Cruise, integrating self-driving cars into its network in select cities. While Uber abandoned its own self-driving ambitions in 2020, Khosrowshahi predicts autonomous vehicles will dominate the platform within 15-20 years, potentially phasing out human drivers entirely.

Source: Entrepreneur

PAPERS, VIDEOS & ARTICLES TO STUDY ??

DeepSeek: The End of the AI Model Race?

DeepSeek’s emergence signals a shift in AI from closed, compute-intensive models to open-source efficiency, potentially ending the dominance of corporate AI races. Experts argue that AI’s future now hinges on widespread adoption, cost-effective inference, and specialized integration rather than sheer computing power. While Nvidia’s grip on AI hardware remains strong, DeepSeek’s approach could challenge Big Tech’s control. The real battleground now lies in data ownership, customization, and seamless AI integration. As AI moves toward efficiency and accessibility, the next competitive advantage won’t be knowing everything—it’ll be knowing where to look.

Source: Insead

China’s EV Giants Pivot to Humanoid Robots

With profit margins shrinking in China’s fiercely competitive EV market, automakers like BYD, XPeng, and Nio are expanding into humanoid robotics. Leveraging their expertise in sensors, AI, and battery technology, these companies are developing cost-effective robots for industrial applications. China’s dominance in the global supply chain for key robotics components gives it a manufacturing edge, with Unitree’s H1 priced at less than half of Boston Dynamics’ Atlas. However, challenges remain in AI and chip development, where Western firms lead. Backed by government subsidies, China’s EV-to-robotics shift mirrors its past success in electric vehicles.

Source: MIT Technology Review

Five AI Trends Leaders Must Watch in 2025

AI experts Thomas H. Davenport and Randy Bean highlight five key trends reshaping business in 2025. Leaders must balance the promise and hype of agentic AI while ensuring measurable ROI on AI investments. Building a true data-driven culture remains a challenge, as does managing the surge of unstructured data. Meanwhile, AI leadership roles and reporting structures continue to evolve. With AI’s rapid advancements, businesses must cut through the noise, focus on strategic implementation, and harness AI responsibly for real impact. Understanding these shifts will be crucial for staying competitive in the AI-driven era.

Source: MIT Sloan Management Review

OTHER NEWS IN BRIEF ??

Meta is working on ‘humanoid’ robots, too Meta is planning to design the hardware and software for humanoid robots, a newly formed team within Meta’s Reality Labs division will start by working on “humanoid robot hardware” capable of completing household chores.

Perplexity just made AI research crazy cheap—what that means for the industry Perplexity's Deep Research tool matches $75,000/month enterprise AI capabilities, forcing OpenAI and Google to justify premium pricing

5 hot takes on reinventing old age as a concept We have reflected long and hard on many subjects, but no one really paid attention to population aging until about the 1930s

DeepSeek founder Liang Wenfeng is reportedly set to meet with China’s Xi Jinping Chinese AI startup DeepSeek founder Liang Wenfeng is reportedly set to meet with China’s top politicians, including Chinese leader Xi Jinping, during a summit that Alibaba founder Jack Ma is also expected to attend

Scarlett Johansson warns of dangers of AI after Kanye West deepfake goes viral Scarlett Johansson has warned of the “imminent dangers of AI” after a deepfake video of her and other prominent Jewish celebrities opposing recent antisemitic remarks from Kanye West went viral this week

Meta’s Ray-Bans have sold 2 million pairs — its maker is prepping to sell 10M each year The company might theoretically reach 2 million or even 5 million by the end of 2025.

Researchers find you don’t need a ton of data to train LLMs for reasoning tasks With a few hundred well-curated examples, an LLM can be trained for complex reasoning tasks that previously required thousands of instances

Too many models, too much confusion: OpenAI pledges to simplify its product line Altman is the rare CEO to admit that the number of models the company released has made their core products more complicated to use

A ‘true crime’ YouTube channel’s videos got millions of views. It turns out the stories were AI-generated A popular “true crime” YouTube channel has been pulling in millions of views with videos about gruesome murders. As it turns out, none of them are real.