May 28, 2025

news  
Read News
European AI Startups Boom, Google Antitrust Concerns, Tesla Autonomy Advances

The AI landscape is rapidly evolving, with significant investments and advancements in various sectors. European AI startups have seen a 55% increase in investment, with the European Commission aiming to set standards for trustworthy AI. Meanwhile, Sheba Medical Center is pioneering AI integration in healthcare, and Google's AI updates have raised antitrust concerns. AI model collapse is a growing concern, and experts warn of its potential consequences. Microsoft has introduced an AI-powered Write feature to Notepad, and OpenAI is developing a new hardware product to extend ChatGPT beyond software. Tesla's AI technology has accumulated over 3.5 billion miles with supervision, and analysts believe its autonomy and robotaxi service could unlock significant valuation. Companies like James & James are using marketing mix modeling to keep AI-based ad buying honest, and Google has introduced a new AI Mode feature to create a more conversational search experience. Independent researchers are also analyzing AI system prompts to understand how companies control their models' behavior.

Key Takeaways

  • European AI startups have seen a 55% increase in investment in Q1 2025.
  • Sheba Medical Center is integrating AI into every aspect of hospital operations to transform healthcare delivery.
  • Google's AI updates have raised antitrust concerns due to the company's dominance in the AI market.
  • AI model collapse is a growing concern that can lead to distorted data distributions and irreversible defects in performance.
  • Microsoft has introduced an AI-powered Write feature to Notepad to generate and refine text using prompts.
  • OpenAI is developing a new hardware product to extend ChatGPT beyond software and run AI locally.
  • Tesla's AI technology has accumulated over 3.5 billion miles with supervision and plans to deploy fully autonomous robotaxis.
  • Companies like James & James are using marketing mix modeling to keep AI-based ad buying honest and effective.
  • Google has introduced a new AI Mode feature to create a more conversational search experience.
  • Independent researchers are analyzing AI system prompts to understand how companies control their models' behavior and advocate for transparency.

Europe Leads AI Startup Growth

European AI startups have seen a 55% increase in investment in Q1 2025, with nearly half of all unicorns now being AI-driven. The European Commission has increased its investment efforts towards AI, aiming to set standards for trustworthy and human-centric AI. A list of 20 promising AI startups founded between 2022 and 2025 has been compiled, including Adaptive ML, Bioptimus, and Black Forest Labs. These startups are working on various AI applications, such as reinforcement learning, biological data integration, and generative deep learning models.

Sheba Medical Center's AI Vision

Sheba Medical Center is pioneering a global vision to integrate AI into every aspect of hospital operations. The center's ARC innovation ecosystem serves as a platform for transforming healthcare delivery locally and globally. Sheba's AI initiatives are integral to its transformation roadmap, aiming to establish the hospital as a global leader in AI-driven healthcare transformation. The center's AI Center will focus on transforming healthcare through multidisciplinary task forces, and collaborations with startups like Aidoc, MDClone, and Nucleai are central to ARC's innovation strategy.

Acing AI Hiring Assessments

Career coach Herman Ko shares five tips for acing AI hiring assessments, including making a strong first impression, thinking like a top 1% candidate, maintaining a professional image, keeping it natural, and not using AI in interviews. Ko emphasizes the importance of being authentic and not trying to cheat the system. He also notes that AI hiring assessments are becoming increasingly common, and candidates need to be prepared to showcase their skills and personality in a virtual setting.

Google's AI Updates Raise Antitrust Concerns

Google's recent AI updates have raised concerns among regulators and industry experts, who worry that the company's dominance in the AI market could lead to antitrust issues. Google's AI-powered advertising products and capabilities, such as Ads in AI Overviews and AI Mode on Search, have the potential to give the company an unfair advantage over its competitors. Experts are advising marketers to carefully evaluate the implications of using Google's AI tools and to consider the potential risks of relying too heavily on a single company's technology.

AI Model Collapse: A Growing Concern

AI model collapse refers to the phenomenon where AI systems, trained on their own outputs, gradually lose accuracy, diversity, and reliability. This can lead to distorted data distributions and irreversible defects in performance. The issue is attributed to three factors: error accumulation, loss of tail data, and feedback loops. Experts warn that AI model collapse can have significant consequences, including the production of misleading or false information, and emphasize the need for careful evaluation and testing of AI systems to mitigate this risk.

Microsoft Notepad Gets AI Makeover

Microsoft is adding an AI-powered Write feature to Notepad, allowing users to generate and refine text using prompts. The feature is part of a Windows 11 update and builds on other AI tools that Notepad has introduced in the past year. The Write feature uses Windows Copilot to produce a prompt box, and users can submit prompts and see what the AI comes up with. The feature aims to make Notepad more active and collaborative, encouraging users to work with Microsoft's AI interfaces.

OpenAI to Ship 100 Million Secret AI Devices

OpenAI is developing a new hardware product that will ship 100 million units, aiming to extend ChatGPT beyond software. The device, which is being developed in collaboration with former Apple chief design officer Jony Ive, is expected to run AI locally, reducing server load and costs. However, the device's success will depend on its ability to provide clear advantages over smartphones and other existing devices.

Tesla's Sales Decline, But Analysts Remain Upbeat on AI Potential

Tesla's sales have declined significantly in Europe, but analysts remain optimistic about the company's long-term prospects due to its artificial intelligence and autonomous driving ambitions. Tesla's AI technology has accumulated over 3.5 billion miles with supervision, and the company plans to deploy fully autonomous robotaxis starting with a modest fleet in Austin by the end of June 2025. Analysts believe that Tesla's autonomy and robotaxi service could unlock approximately $1 trillion in AI-related valuation, potentially propelling the company's market capitalization to $2 trillion by late 2026.

James & James Uses MMM to Keep AI-Based Ad Buying Honest

James & James, a furniture brand, is using marketing mix modeling (MMM) to keep its AI-based ad buying honest. The company's CMO, Tristan Cameron, worked with Inuvo to create a quarterly MMM benchmark, which revealed that the brand's AI-based ad products were not pulling their weight. The benchmark helped James & James to pull the plug on its Advantage+ Shopping Campaigns budget and to shift its focus to manually controlled ad products. The company's experience highlights the importance of careful oversight and evaluation of AI-based ad products to ensure that they are effective and efficient.

Google's New AI Mode Feature

Google has introduced a new AI Mode feature that allows users to ask follow-up questions and receive comprehensive answers. The feature is part of Google's efforts to integrate AI into its search products and to create a more conversational search experience. However, some critics worry that the introduction of AI Mode could signal the end of the open web, as users may rely more heavily on Google's AI-powered search results rather than clicking on external links. Google's CEO, Sundar Pichai, has emphasized that the company is committed to expanding the web, rather than replacing it.

Hidden AI Instructions Reveal How Anthropic Controls Claude 4

Independent AI researcher Simon Willison has published an analysis of Anthropic's system prompts for Claude 4, revealing how the company controls the model's behavior through its outputs. The analysis shows that Anthropic includes instructions for the model to provide emotional support, avoid encouraging self-destructive behavior, and use only one short quote from web sources per response. Willison's findings provide valuable insights into how Anthropic's AI models work and how they can be used effectively. He also calls on Anthropic and other AI companies to be more transparent about their system prompts and to publish them officially.

Sources

AI Artificial Intelligence Machine Learning Deep Learning European AI Startups AI in Healthcare