SOBOTA 22.2.2025: NAJNOVŠIE AI SPRÁVY ZA POSLEDNÝCH 24 HODÍNSATURDAY, FEBRUARY 22, 2025: LATEST AI NEWS FROM THE PAST 24 HOURS

If you missed yesterday’s (February 21, 2025) flood of news about artificial intelligence, don’t worry! We bring you a comprehensive and viral overview of the most important AI news from the last 24 hours-from remarkable research to unexpected blunders. Sit back, relax, and let’s get started!

I. TOP AI NEWS OF THE DAY: OPENAI’S “OPERATOR” IS GOING GLOBAL

Why is this a big deal? OpenAI is expanding its AI agent “Operator”-a tool that can act on your behalf in online environments, clicking links and filling out forms. Until now, it was only available with the more expensive ChatGPT Pro subscription in the US, but as of yesterday, other countries like Canada, Brazil, Japan, and Singapore have been added. All of this excludes the EU, which has been left out (for now) due to stricter regulations.

User experience: The agent runs in a separate browser window; you control it via voice or text, and it performs (almost) all online tasks for you.

Security: For now, the AI won’t fill in sensitive data like passwords without your explicit consent, so this doesn’t mean unrestricted access to credit cards.

Expectations: Tech analysts say that AI assistants like Operator are a major step toward even more autonomous virtual agents that will save people hours of time each week.

II. TOP AI TRENDS AND DEVELOPMENTS

1) The xAI Controversy: Does It Dare to Suggest the Death Penalty?

Elon Musk’s company xAI is grappling with a curious fiasco: its chatbot Grok initially spouted names like Donald Trump or even Elon Musk himself in response to odd questions about “who deserves the death penalty in the U.S.” Of course, xAI quickly issued a correction and called it a “really terrible failure.” Grok now responds to similar questions with a typical evasive statement.

2) Chinese AI lab DeepSeek continues its open-source push

DeepSeek surprised everyone a month ago by making the R1 model freely available. Now it promises to publish five main code repositories that are already in live operation. Apparently, they want the community to join in and further develop them. The contrast with the big Western giants (Meta, OpenAI) couldn’t be greater.

3) Apple is integrating Apple Intelligence into Vision Pro

The Vision Pro headset, which Apple positions as a “spatial computer,” is getting AI features. For example, text and image generation, as well as the ability to summarize messages. iOS 18.4 also includes a new feature called Priority Notifications-that is, intelligent prioritization of notifications. Further proof that Apple is serious about AI and promises that this is supposedly “just the beginning.”

4) Meta and the Copyright Controversy

Court documents state that Meta employees repeatedly discussed the use of libraries like Libgen in internal chats-even though they suspected the content was copyrighted. According to leaked reports, they were even aware that it was legally “questionable,” but argued “that others are doing it too.” The lawsuit is ongoing and raises the question of whether collecting copyrighted content for AI models still falls under “fair use.”

5) Google’s AI Co-Scientist

Google has announced an advanced AI that helps scientists formulate hypotheses, design experiments, and accelerate scientific discoveries. The project has reportedly already helped in real-world laboratory settings-for example, in research on antibiotic resistance or in the search for new ways to degrade plastics using enzymes.

III. BREAKTHROUGH AI APPLICATIONS

1) AI-Assisted Plastic-Digesting Enzymes

Another milestone: Researchers at Google AI used the generative models RFDiffusion and PLACER to design artificial enzymes capable of breaking down plastic polymers. This could herald a new era in recycling and eco-friendly waste processing.

2) Humanoid robots for the home

Startups Clone Robotics and 1X from Norway have unveiled new prototypes of AI-powered humanoid robots capable of hanging from the ceiling, moving their “muscles,” and even doing laundry. Autonomous operation is still a long way off, but test models suggest that the average household could welcome its first “iron helper” within a few years.

3) Figure AI and Helix

Figure AI has introduced the Helix model, which combines vision, language comprehension, and motor skills so that humanoid robots can perform nearly “human-like” tasks, react to new objects, and collaborate with one another. According to demonstrations, two robotic assistants equipped with Helix can organize groceries in the kitchen or safely handle unfamiliar objects.

IV. THE IMPACT OF AI ON BUSINESS AND DAILY LIFE

1) Nvidia and Its Sign Language

Nvidia’s Signs platform expands the possibilities for teaching American Sign Language (ASL). Using a 3D avatar and a camera, the AI evaluates gestures and provides the user with immediate feedback. The goal is to support parents of deaf children and create an open database for further AI applications.

2) EV Charging Station Ecosystem and AI

Research suggests that electric power grids can more effectively integrate artificial intelligence to identify electric vehicle charging and better manage consumption peaks. AI would recognize in real time which drivers need to charge urgently and where this places the greatest strain on the distribution network.

3) Billions of Dollars at Stake: Investments in AI Startups

Elon Musk’s xAI Grok 3-promising a revolution in reasoning capabilities and fierce competition with ChatGPT.

Genspark raised $100 million, with the ambition to challenge Google in the field of AI search.

Together AI raised $305 million for its AI cloud, enabling developers to easily deploy both open-source and proprietary models.

UK-based Humanoid unveiled its new series of robots, the HMND 01.

V. GLOBAL TRENDS IN AI

1) France and South Korea

French President Macron announced massive investments in AI infrastructure and data centers (over 100+ billion euros) with an emphasis on “clean energy” for AI computing.

South Korea plans to build the world’s largest AI data center, collaborating with LG and an investment consortium.

2) AI and Artistic Rights

A heated debate erupted at Christie’s in London when AI-generated works were put up for auction. Artists are protesting that such creations use illegitimately copied styles. Christie’s claims that most of the works were trained on their own source material, but the discussion about the ethics of AI art continues.

3) Risks and Regulations

With the rise of AI come cybersecurity concerns-in particular, the threat of AI-driven fraud and phishing is coming to the forefront. Studies suggest that companies generally do not have a plan in place for generative AI security. Furthermore, many governments are tightening oversight (EU, South Korea).

VI. PROPOSALS FOR PROCESS OPTIMIZATION AND DIGITALIZATION WITH AI

Implement AI into ERP and CRM systems: automated processing of inquiries, invoicing, rapid risk analysis.

Use Operator-type agents (OpenAI) for routine tasks: bookings, searches, writing emails.

Use AI models for market research (e.g., sentiment analysis or monitoring competitors on social media).

Test your own small language models (e.g., DistilBERT) for fast and inexpensive text analysis.

Train employees on AI: so they aren’t afraid to use AI tools and understand the risks (e.g., copyright).

THREE QUESTIONS AND ANSWERS THAT CAN MAKE YOUR LIFE EASIER

Question: How can you quickly integrate an AI agent (such as Operator) into business processes without compromising data security?Answer:

First, conduct a pilot project in a department where the risk of sensitive information leaks is low (e.g., HR for processing resumes).

Ensure that the AI agent does not have access to internal databases without human validation.

Establish clear rules and protocols: who can grant the AI agent access to sensitive passwords and when.

Question: I need a state-of-the-art AI model, but I’m concerned about gray areas in copyright law. How do I choose a safe strategy?Answer:

• Audit the data source find out where the model draws its text from and whether licenses exist.
• Use open-source models (e.g., those published by DeepSeek with transparent code) so you know what the model contains.
• If you’re planning commercial use, consult lawyers regarding “fair use” and any applicable licensing agreements.
• Question How can I effectively invest in AI infrastructure if I’m a medium-sized company without massive capital?Answer:
• Use cloud solutions like AWS, Azure, or startups like Together AI, which offer flexible packages.
• Deploy smaller/more compact models (DistilBERT or Llama 7B) and only switch to more robust solutions when expanding projects.
• Look for strategic investors or VC funds that have special programs for AI.

CONCLUSION

We see that yesterday (February 21, 2025) was another crazy day full of news: AI agents, humanoid robots, copyright battles, and plastic-degrading enzymes. The world of artificial intelligence is advancing at an unstoppable pace, and this hasn’t come without controversies (authorship of artistic works), minor scandals (a chatbot suggesting death penalties), or massive investments in development (Grok 3, Genspark, Together AI).

Be sure to turn on notifications for Inovator AI.com and follow us, because more groundbreaking changes could come as early as tomorrow. This field is too dynamic to ignore-and at the same time, too exciting to let ourselves get carried away by its potential. See you at the next “AI Update”!

(Below is a complete list of all international sources used and their links from the attached material, sorted by serial number.)

Ars Technica AI – https://arstechnica.com/ai/2025/02/deepseek-goes-beyond-open-weights-ai-with-plans-for-source-code-release/

Ars Technica AI – https://arstechnica.com/gadgets/2025/02/dangling-twitching-human-robot-with-synthetic-muscles-makes-its-debut/

The Verge AI – https://www.theverge.com/news/617799/elon-musk-grok-ai-donald-trump-death-penalty

The Verge AI – https://www.theverge.com/23610427/chatbots-chatgpt-new-bing-google-bard-conversational-ai

The Verge AI – https://www.theverge.com/24246632/apple-intelligence-ios-18-ipad-os-18-macos-sequoia-iphone-16

The Verge AI – https://www.theverge.com/the-vergecast/617137/iphone-16e-humane-ai-pin-vergecast

AI News – https://www.ainews.com/p/google-s-ai-co-scientist-how-ai-is-accelerating-scientific-discovery

AI News – https://www.ainews.com/p/google-scientists-use-ai-to-design-a-plastic-digesting-enzyme

AI News – https://www.ainews.com/p/figure-ai-s-helix-enables-robots-to-see-understand-and-act-like-humans

AI News – https://www.ainews.com/p/nvidia-s-ai-powered-signs-platform-expands-asl-learning-accessibility

AI Business – https://aibusiness.com/automation/ai-driven-wireless-tree-network-connects-wildfire-defense-system

AI Business – https://aibusiness.com/generative-ai/most-read-elon-musk-xai-unveils-grok-3-to-challenge-rivals-world-s-largest-ai-data-center-planned-for-south-korea

AI Business – https://aibusiness.com/automation/ai-startup-humanoid-unveils-general-purpose-humanoid-robot

AI Business – https://aibusiness.com/automation/ai-driven-wireless-tree-network-connects-wildfire-defense-system

KDnuggets – https://www.kdnuggets.com/distilbert-resource-efficient-natural-language-processing

AI Magazine – https://aimagazine.com/articles/this-weeks-top-5-stories-in-ai-21-february-2025

TechCrunch AI – https://techcrunch.com/2025/02/21/openai-rolls-out-its-ai-agent-operator-in-several-countries/

TechCrunch AI – https://techcrunch.com/2025/02/21/techcrunch-disrupt-2025-lowest-prices-of-the-year-end-in-7-days/

TechCrunch AI – https://techcrunch.com/2025/02/21/deepseek-to-open-source-parts-of-online-services-code/

TechCrunch AI – https://techcrunch.com/2025/02/21/apply-to-speak-at-techcrunch-sessions-ai-before-the-deadline/

TechCrunch AI – https://techcrunch.com/2025/02/21/report-ai-coding-assistants-arent-a-panacea/

TechCrunch AI – https://techcrunch.com/podcast/ai-wearables-1-0-was-humanes-ai-pin-too-ambitious/

TechCrunch AI – https://techcrunch.com/2025/02/21/the-vision-pro-is-getting-apple-intelligence-in-april/

TechCrunch AI – https://techcrunch.com/2025/02/21/sakana-walks-back-claims-that-its-ai-can-dramatically-speed-up-model-training/

TechCrunch AI – https://techcrunch.com/2025/02/21/norways-1x-is-building-a-humanoid-robot-for-the-home/

TechCrunch AI – https://techcrunch.com/2025/02/21/report-openai-plans-to-shift-compute-needs-from-microsoft-to-softbank/

TechCrunch AI – https://techcrunch.com/2025/02/21/nvidia-ceo-jensen-huang-says-market-got-it-wrong-about-deepseeks-impact/

TechCrunch AI – https://techcrunch.com/2025/02/21/ios-18-4-will-bring-apple-intelligence-powered-priority-notifications/

TechCrunch AI – https://techcrunch.com/2025/02/21/court-filings-show-meta-staffers-discussed-using-copyrighted-content-for-ai-training/

IBM Research Blog – https://research.ibm.com/blog/gneissweb-for-granite-training

ElevenLabs Blog – https://elevenlabs.io/blog (specifically the article on Impact Program V2 dated February 21, 2025)

Alibaba Cloud AI Blog – https://www.alibabacloud.com/blog/alibabas-core-businesses-reignite-growth-as-ai-strategy-delivers-strong-results_602006