The News Feed

Blog

  • Perplexity AI Offers to Merge with TikTok’s US Unit

    Perplexity AI Offers to Merge with TikTok’s US Unit

    In a surprising turn of events, the U.S.-based AI search engine startup, Perplexity AI, has proposed a merger with TikTok’s U.S. operations. This bold move comes as TikTok faces the possibility of being banned in the United States due to national security concerns tied to its Chinese parent company, ByteDance.

    Details of the Bid

    Merger Proposal

    Perplexity AI’s plan involves creating a new entity by merging its operations with TikTok U.S. and bringing in new capital partners. This approach allows ByteDance’s current investors to keep their equity stakes, offering a compromise that could be more acceptable to ByteDance compared to an outright sale.

    Rationale

    The merger would add TikTok’s massive video content library and user base to Perplexity’s AI-powered search platform. This integration could enhance Perplexity’s offerings and bring new ways for users to discover content.

    Timing

    The bid comes at a critical time, just before a U.S. law requiring ByteDance to sell TikTok’s U.S. operations or face a ban is set to take effect on Sunday, January 19.

    Context and Implications

    TikTok’s Situation

    The urgency stems from a recent Supreme Court ruling that supports the law mandating ByteDance to divest TikTok’s U.S. assets. With President Joe Biden nearing the end of his term, enforcement will be left to the incoming Trump administration.

    Perplexity AI’s Strategy

    Perplexity AI, known for its innovative AI search technology, aims to expand its influence in the digital world by combining with TikTok, one of the most popular social media platforms globally. This merger could pave the way for new applications of AI in creating and sharing content.

    Market Reaction

    The news has sparked debates online. While some view it as a creative way to preserve TikTok’s U.S. presence, others question how the merger would address data privacy concerns and control issues.

    Next Steps

    Trump’s Influence

    President-elect Donald Trump has hinted at the possibility of granting TikTok a 90-day extension, allowing more time for negotiations or additional bids.

    ByteDance’s Stance

    ByteDance has historically opposed selling TikTok, considering it a key part of its business. However, a merger might present an alternative to a complete sale, which could be more appealing.

    Sources

    Reports from CNBC, Reuters, and Bloomberg confirm the proposal and provide insights into the intentions behind the merger plan.

    This potential deal represents a pivotal moment for TikTok, Perplexity AI, and the future of social media in the U.S.

  • Amazon Temporarily Suspends  Drone Deliveries After Crashes

    Amazon Temporarily Suspends  Drone Deliveries After Crashes

    Amazon has stopped its drone delivery service in the United States for now. This decision comes after some crashes during testing at a facility in Pendleton, Oregon.

    Details of the Incidents

    In December, two drones, called the MK30 models, crashed in rainy weather. The light rain caused a software issue, which led to the accidents. One of the drones even caught fire after crashing.

    This is not the first challenge for Amazon’s Prime Air drone program, which has been in development for more than ten years.

    Operational Pause

    Because of these incidents, Amazon has paused its drone deliveries in Texas and Arizona. The company is working on fixing the software problems.

    Av Zammit, an Amazon spokesperson, explained, “The purpose of these tests is to push our aircraft past their limits – it would be irresponsible not to do that. We are committed to ensuring the safety of our operations, and this pause will allow us to do just that.”

    Background and Impact

    Amazon’s Prime Air program aims to deliver packages in 30 minutes or less using drones. Last year, the company got approval from the Federal Aviation Administration (FAA) to fly drones beyond the operator’s line of sight.

    However, the program has faced many challenges, including a 2022 crash that caused a 25-acre brush fire. This suspension adds to concerns about the safety and readiness of drone deliveries.

    Broader Implications

    The delay affects Amazon’s goal to deliver 500 million packages by drone every year by the end of this decade. It also raises questions about whether drones are ready for widespread use in deliveries.

    Looking Forward

    Amazon says it remains committed to drone delivery. The company has recently expanded operations to Italy and plans to start in the UK soon.

    For now, Amazon is fixing the software problems and waiting for the FAA to approve the u

  • Perplexity Buys Read.cv, a Social Media Platform, to Boost Professional Networking

    Perplexity Buys Read.cv, a Social Media Platform, to Boost Professional Networking

    In a major move to grow its presence in professional networking, AI-driven search engine Perplexity has announced it is buying Read.cv. The announcement was made on January 17, 2025, and shows Perplexity’s commitment to improving user connections and making knowledge more accessible.

    End of an Era for Read.cv

    Read.cv was known as a simple and creative platform for professionals to share their work, projects, and achievements. Starting today, the platform will begin shutting down. Users have until May 16, 2025, to save their data, including profiles and messages, using a built-in data export tool. The team at Read.cv described the closure as “bittersweet,” reflecting on their platform’s focus on creativity and curiosity.

    New Beginnings with Perplexity

    Both companies view the acquisition as a perfect fit for their goals. Read.cv shared in a farewell blog post, “We’ve long admired Perplexity and believe great things happen when the world’s knowledge is made more open and accessible.” Perplexity’s CEO, Aravind Srinivas, expressed excitement about welcoming Read.cv’s design and engineering teams to work on shared goals of exploration and innovation using AI.

    Backed by investors like Jeff Bezos and NVIDIA, Perplexity aims to combine AI with social innovation. This acquisition could help the company create new ways for professionals to connect and collaborate.

    Reactions and Future Outlook

    The news has brought mixed feelings from Read.cv users and the tech world. Many feel sad to see the platform close, but some are hopeful about the new possibilities. “It’s sad to see Read.cv sunset, but the merger with Perplexity could bring exciting new features for professionals,” commented one user online.

    While Perplexity has not shared exact plans for how Read.cv’s features will be used, they have hinted at new ideas for improving professional networking with AI tools.

    About the Companies

    Perplexity: An AI-powered search engine focused on providing accurate answers and valued at $9 billion. It is supported by major investors like Jeff Bezos and NVIDIA.

    Read.cv: A platform created by Andy Chung, a former designer at Facebook and Mozilla, known for emphasizing creativity in professional profiles.

    This acquisition highlights the growing importance of AI and user-focused design in shaping how professionals connect and share online.

  • Google Adds AI to Workspace and Increases Pricing

    Google Adds AI to Workspace and Increases Pricing

    Google has made a big update to its Google Workspace tools. The company is now adding its advanced AI tool, Gemini, to all Business and Enterprise plans. Along with this, the prices for these plans are going up. This change shows Google’s focus on making AI a key part of its productivity tools, like Gmail, Docs, and Meet.

    Price Adjustments

    The new prices are already in effect for new customers. Existing customers will see the price changes starting March 17, 2025. Here’s the updated pricing:

    • Business Starter: $6 → $7 per user per month (annual basis)
    • Business Standard: $12 → $14 per user per month (annual basis)
    • Business Plus: $18 → $22 per user per month (annual basis)

    Google says the higher prices reflect the cost of adding AI features directly into the plans without charging extra for add-ons.

    Gemini AI Integration

    The Gemini AI tool is now included in all paid Workspace plans. Before this, users needed to buy separate AI add-ons, like Gemini Business for $20 per user per month or Gemini Enterprise for $30 per user per month. Now, these features are part of the standard subscription.

    With Gemini AI, users can:

    • Take automatic notes during Meet calls.
    • Create images in Docs and Slides.
    • Use improved tools like NotebookLM.

    Other AI-powered tasks include summarizing emails, writing assistance, and real-time transcription. Google aims to make these tools accessible without extra costs.

    Reactions and Implications

    Not everyone is happy about the changes. Some users appreciate the value of AI features for a slightly higher price. Others feel it’s unfair, especially if they don’t need the AI tools but still have to pay more.

    Social media reactions have been mixed. While some are excited about using AI in their daily work, others are frustrated about paying more for features they might not use.

    Future Outlook

    This update highlights a growing trend of AI becoming a regular part of office tools. Google’s move could push competitors, like Microsoft, to include AI in their base plans or find new ways to charge for similar features.

    Google says the added value of Gemini AI should make up for the price increase for most users. The company hopes to make AI a standard part of the workplace, not just an optional extra.

  • Apple Stops AI-Generated News Summaries After Generating  Fake Headlines

    Apple Stops AI-Generated News Summaries After Generating  Fake Headlines

    Apple has temporarily stopped using its AI-powered feature that creates summaries for news and entertainment notifications. This decision comes after the technology made errors, leading to the spread of incorrect information.

    What Happened?

    The issue began when Apple’s AI misrepresented a news story. For example, a BBC article about a high-profile murder case was incorrectly summarized. The AI summary falsely claimed that Luigi Mangione, a person involved in the case, had taken his own life. However, this was not true according to the actual news.

    Changes in the Latest Update

    Apple has made several changes with the release of iOS 18.3, iPadOS 18.3, and macOS Sequoia 15.3 beta updates:

    • Italicized Text: AI-generated summaries will now appear in italics, so users can easily identify them.
    • User Control: Users can now turn off these summaries for specific apps directly from their Lock Screen.
    • Beta Warning: A new message warns users that the feature is still in beta and might not always be accurate.

    Apple’s Plan for Improvement

    Apple has promised to fix the issues and bring back the feature when it is more reliable. This is part of its larger effort to improve its AI system, known as Apple Intelligence.

    Why It Matters

    This incident shows the challenges of using AI for important tasks like summarizing news. Mistakes in such cases can lead to misinformation. By pausing the feature, Apple is showing responsibility in addressing these challenges and ensuring better accuracy in the future.

    Broader Implications

    Apple’s decision has started a bigger conversation about the role of AI in news, the risks of misinformation, and the importance of transparency when using AI in everyday products.

  • Mistral Partners with AFP to Offer Up-to-Date Answers in Le Chat

    Mistral Partners with AFP to Offer Up-to-Date Answers in Le Chat

    In a big step for the AI world, French startup Mistral has signed a multi-year deal with Agence France-Presse (AFP), one of the top news agencies globally. This agreement will allow Mistral’s AI chatbot, Le Chat, to use AFP’s huge news archive to give users more accurate and up-to-date answers.

    A Key First Partnership for Mistral

    The deal, announced soon after Google’s partnership with The Associated Press, is Mistral’s first with a major news organization. AFP will share its daily output of around 2,300 news stories in six languages: Arabic, English, French, German, Portuguese, and Spanish. In addition, Le Chat will have access to AFP’s archives going back to 1983. However, the agreement does not include AFP’s photos or videos.

    Improving AI Accuracy

    Arthur Mensch, Mistral’s CEO and co-founder, highlighted the importance of this deal. He said, “We believe improving the accuracy of [Le Chat’s] responses is a key step in the deployment of our technology, particularly for businesses.” The partnership will help make Le Chat more reliable, especially for companies that need accurate information in multiple languages.

    Benefits for AFP

    For AFP, this collaboration brings financial advantages at a time when traditional media is facing challenges due to digital changes. AFP’s CEO, Fabrice Fries, stated, “Through this partnership, AFP is further diversifying its revenue sources.”

    Addressing AI Challenges

    The partnership comes as companies look for ways to make AI systems more trustworthy. By using verified news from AFP, Le Chat aims to reduce errors or “hallucinations” – when AI generates incorrect information. This approach could become a model for how AI and content providers work together while addressing copyright concerns.

    Rolling Out Soon

    AFP’s content will be integrated into Le Chat in the coming weeks. Users can expect a more fact-based, reliable experience, thanks to this collaboration. This move is part of Mistral’s plan to offer practical and trustworthy AI solutions.

    Changing How We Access News

    This partnership is sparking discussions in the tech and journalism world. While many see this as a way for AI to transform news consumption, the real test will be whether Le Chat can present the information in a user-friendly and accurate way.

    As more AI companies partner with traditional media, the way we access and trust news is set to evolve, giving people diverse and reliable sources through advanced AI tools.

    Links

    https://mistral.ai/news/mistral-afp

  • Microsoft Launches AI-Powered Copilot Chat with Pay-as-You-Go Pricing

    Microsoft Launches AI-Powered Copilot Chat with Pay-as-You-Go Pricing

    Microsoft has introduced a new addition to its Microsoft 365 suite called Microsoft 365 Copilot Chat, launched on January 15, 2025. This tool is aimed at businesses and comes with a flexible pay-as-you-go model for AI agents, making it a practical option for everyday business operations.

    New Features and Pricing Model

    Microsoft 365 Copilot Chat uses OpenAI’s latest model, GPT-4o, to provide a secure and simple AI chat experience. Basic use of the chat feature is free for all Microsoft 365 commercial customers.

    The standout feature is the consumption-based pricing model for AI agents, which allows businesses to automate repetitive tasks like customer service and internal processes. Here’s the pricing breakdown:

    • Basic Answers: 1 message per response, costing $0.01 per message.
    • Generative Answers: 2 messages per response, also $0.01 per message.
    • Autonomous Actions: 25 messages per action, costing $0.25 per action.

    This model provides a cost-effective alternative to the fixed $30 per user per month subscription for the full Microsoft 365 Copilot.

    Enterprise Data Protection and IT Controls

    Microsoft has focused on ensuring that Copilot Chat is secure and manageable for businesses by including:

    • Enterprise Data Protection (EDP): This feature safeguards company data while using AI.
    • IT Admin Controls: These controls allow businesses to manage agent deployment, access, and usage effectively.

    Strategic Implications

    The pay-as-you-go model is part of Microsoft’s strategy to make AI more accessible and practical for businesses of all sizes. By allowing companies to scale their AI usage based on demand, Microsoft aims to address concerns about the high cost and complexity of adopting AI.

    This move encourages businesses to integrate AI into workflows, enabling employees in customer service, marketing, and technical roles to use AI tools with ease.

    User and Industry Response

    The launch has sparked interest and excitement, with many businesses seeing it as an opportunity to adopt AI without committing to high-cost subscriptions. However, some small and medium-sized enterprises are skeptical about the cost-effectiveness of the model, as usage costs may vary.

    Industry analysts believe this move will put pressure on competitors to introduce flexible pricing or enhance their AI offerings. The integration of AI agents into daily tasks could lead to more efficient and autonomous workflows.

    Looking Ahead

    Microsoft plans to continue expanding Copilot’s capabilities with more sophisticated AI features and deeper integration across Microsoft 365 apps. The company sees AI as a critical tool for the future of business, and this launch is a significant step toward making that vision a reality.

    Links

    https://www.microsoft.com/en-us/microsoft-copilot/blog/copilot-studio/enabling-agents-in-microsoft-365-copilot-chat/

  • NVIDIA Launches Tools to Help Enterprises Adopt AI Agents Safely

    NVIDIA Launches Tools to Help Enterprises Adopt AI Agents Safely

    NVIDIA Corporation has introduced three new tools to help businesses use AI more safely and securely. These tools, called NVIDIA Inference Microservices (NIM), are designed to ensure AI systems follow rules and stay ethical. They are part of NVIDIA’s plan to encourage more companies to adopt AI technology.

    What Are the New Tools?

    The new tools are part of NVIDIA’s NeMo Guardrails framework, which helps developers build trustworthy AI systems. The three tools focus on:

    • Content Safety: This tool stops AI systems from creating harmful or biased content. It ensures responses meet ethical standards.
    • Topic Control: This service keeps AI conversations on-topic and avoids inappropriate discussions.
    • Jailbreak Detection: This feature protects AI systems from being misused by detecting and blocking attempts to bypass restrictions.

    Why Are These Tools Important?

    More businesses are using AI systems, but concerns about safety and trust slow down adoption. NVIDIA’s new tools address these concerns, making it easier for companies to use AI agents, or “knowledge robots,” without fear of mistakes or misuse.

    Studies show that around 25% of companies are already using AI agents or plan to start in 2025. By 2027, half of all businesses are expected to use them. NVIDIA’s tools could help speed up this shift by building trust in AI technology.

    A Smarter Way to Use AI

    NVIDIA’s Vice President, Kari Briski, explained why these tools matter:
    “By using small, specialized models, developers can fix gaps that general protections might miss. This approach ensures AI systems are safe and efficient, even in complex situations.”

    These tools also work quickly, even on computers with limited resources. This makes them ideal for companies wanting to use AI on a large scale.

    Teaming Up for Success

    NVIDIA is working with companies like Accenture, Cisco, and SoftServe to bring these tools to more businesses. These partnerships aim to make AI easier to use and safer for everyone.

    Social media platforms like X have seen positive reactions, with many calling this a big step forward for safer AI in business. However, experts note that the true value of these tools will only be clear once they are tested in real-world settings.

    NVIDIA’s Role in Shaping the Future of AI

    NVIDIA is already known for its AI hardware, but this move shows its commitment to solving real-world problems with software too. By making AI safer and more practical, NVIDIA is helping shape how businesses use AI in the future.

    The tech world will be watching closely to see how these tools perform and how they might change the way companies work with AI in the years to come.

  • Chinese AI Startup MiniMax Releases Three Powerful AI Models, Claims Industry-Leading Performance

    Chinese AI Startup MiniMax Releases Three Powerful AI Models, Claims Industry-Leading Performance

    Chinese AI company MiniMax has introduced three advanced AI models, claiming they are competitive with leading technologies from U.S. firms like OpenAI, Google, and Anthropic. Supported by Alibaba and Tencent, MiniMax is making strides in the global AI race.

    MiniMax-Text-01: A Leader in Text Processing, Claims Industry-Leading Performance

    MiniMax-Text-01 is a text-only model with 456 billion parameters, making it one of the largest in its field. It excels at solving math problems and answering fact-based questions, claiming to outperform Google’s Gemini 2.0 Flash in benchmarks like MMLU and SimpleQA. This model can process up to 4 million tokens of text at once, allowing it to analyze massive amounts of information.

    MiniMax-VL-01: Excelling in Multimodal Understanding, Competitive with Top Models

    MiniMax-VL-01 focuses on understanding both text and images. It competes with Anthropic’s Claude 3.5 Sonnet in tasks like interpreting graphs and diagrams, performing well in tests such as ChartQA. While it doesn’t lead in every category, MiniMax claims it holds its ground against other multimodal AI models.

    T2A-01-HD: Advanced Audio Generation, Matching Industry Standards

    T2A-01-HD specializes in audio tasks, including generating speech in 17 languages and cloning voices from short audio clips. Although detailed comparisons are unavailable, the model is said to rival leading audio-generation technologies from Meta and other firms, claiming competitive quality.

    Limited Availability and Legal Issues Challenge Claims

    The new models are available for download on platforms like GitHub and Hugging Face, but they are not fully open-source. The training data and code remain undisclosed, limiting developers’ ability to recreate the models.

    MiniMax is also facing legal challenges. iQiyi, a Chinese streaming platform, claims that MiniMax trained its models using copyrighted content. Additionally, there are reports of the models reproducing logos of British TV channels, raising concerns about copyright violations.

    Global Implications and Challenges for Competitive Claims

    This launch occurs as the U.S. increases restrictions on exporting advanced AI chips to China. These policies, aimed at limiting China’s tech capabilities, could hinder future AI advancements by companies like MiniMax.

    Mixed Reactions from the AI Community on Competitive Position

    Reactions to MiniMax’s announcement are divided. While many praise the technical achievements and competitive claims, others worry about sustainability under legal and political pressures. The ongoing disputes highlight the challenges of navigating international tech competition and intellectual property laws.

    MiniMax’s new models signal China’s determination to compete in the global AI market. However, their success will depend on real-world performance and the company’s ability to address legal and political hurdles.

    Links

    https://minimaxi.com/en/news/minimax-01-series-2

    https://minimaxi.com/en/news/minimax-01-series-2

    https://minimaxi.com/en/news/speech-01-hd-release

  • François Chollet Launches New AI Lab to Pursue Advanced AI Goals

    François Chollet Launches New AI Lab to Pursue Advanced AI Goals

    François Chollet, the creator of the popular Keras deep-learning library, has started a new AI lab called Ndea. This lab aims to create Artificial General Intelligence (AGI), a type of AI that can perform many cognitive tasks at or beyond human ability.

    Ndea’s Mission and Founders

    Chollet co-founded Ndea with Mike Knoop, a former co-founder of Zapier. The lab’s goal is to combine program synthesis with deep learning to build AI that can think, adapt, and innovate like humans. Their work seeks to move beyond today’s AI systems, which are mainly good at specific tasks or recognizing patterns.

    Why Chollet Left Google

    Chollet, who spent almost 10 years at Google, left to focus on Ndea. He has often talked about the limits of large language models (LLMs), which require massive datasets and struggle to generalize from small examples. At Ndea, the team believes program synthesis could be the key to creating more adaptable and intelligent AI.

    Community Reactions to Ndea

    The announcement of Ndea has excited many in the tech world. Posts on X (formerly Twitter) show people discussing Chollet’s expertise and the potential of his ideas. Some are optimistic about Ndea’s approach, while others note the challenges of computational requirements for program synthesis.

    Focus Beyond AGI

    Ndea is not just trying to build AGI but also hopes to study and simulate how humans learn and develop. This could help in fields like drug discovery or materials science, where advanced AI could speed up research.

    Job Openings and Nonprofit Plans

    Ndea is hiring remote researchers and appears to have secured funding, though the details are not public. The company is also starting a nonprofit to create benchmarks for AGI, ensuring transparency and measurable progress in this challenging field.

    A Growing Trend Among AI Researchers

    Chollet’s decision to start his own lab is part of a larger trend. Other well-known researchers, like Ilya Sutskever and Fei-Fei Li, have also launched independent labs. These smaller organizations often explore innovative ideas outside the boundaries of big tech companies.

    Looking Ahead

    Ndea’s journey to create AGI will be closely watched by the AI community. While the challenges are significant, breakthroughs in AGI could redefine what AI can achieve. Many hope this effort will lead to transformative progress in technology and science.

    Links