Elon Musk has echoed concerns from AI researchers that the industry is running out of new, real-world data to train advanced models. Speaking during a livestream with Stagwell’s Mark Penn, Musk noted that AI systems have already processed most of the available human knowledge. He described this data plateau as having been reached last year.
To address the issue, AI developers are increasingly turning to synthetic data, information generated by the AI itself, to continue training models. Musk argued that self-generated data will allow AI systems to improve through self-learning, with major players like Microsoft, Google, and Meta already incorporating this approach in their AI models.
While synthetic data offers cost-saving advantages, it also poses risks. Some experts warn it could cause “model collapse,” reducing creativity and reinforcing biases if the AI reproduces flawed patterns from earlier training data. As the AI sector pivots towards self-generated training material, the challenge lies in balancing innovation with reliability.
Elon Musk’s AI company, xAI, has launched a standalone iOS app for its chatbot, Grok, marking a major expansion beyond its initial availability to X users. The app is now live in several countries, including the US, Australia, and India, allowing users to access the chatbot directly on their iPhones.
The Grok app offers features such as real-time data retrieval from the web and X, text rewriting, summarising long content, and even generating images from text prompts. xAI highlights Grok’s ability to create photorealistic images with minimal restrictions, including the use of public figures and copyrighted material.
In addition to the app, xAI is working on a dedicated website, Grok.com, which will soon make the chatbot available on browsers. Initially limited to X’s paying subscribers, Grok rolled out a free version in November and made it accessible to all users earlier this month. The launch marks a notable push by xAI to establish Grok as a versatile, widely available AI assistant.
The Dutch government announced a deal with Nvidia on Thursday to provide hardware and expertise for a potential AI supercomputing facility. The planned facility is part of the Netherlands‘ broader strategy to bolster AI research and contribute to EU efforts to strengthen Europe’s digital economy.
Last year, the Netherlands allocated €204.5 million ($210 million) for AI investments, with additional funding expected from European subsidies. Economy Minister Dirk Beljaarts hailed the Nvidia agreement as a major step toward realising the project, emphasising the intense global competition for advanced AI technologies.
‘This deal brings building a Dutch AI facility a lot closer,’ Beljaarts said after meeting Nvidia representatives in Silicon Valley, although he refrained from disclosing specific details of the agreement.
Panasonic has introduced Umi, a digital assistant designed to support family wellness, at CES 2025 in Las Vegas. Developed in partnership with AI startup Anthropic, Umi uses the Claude AI model to help families set and achieve personal goals, such as improving fitness or spending more time together. The interactive platform allows users to engage with the AI through voice chat and a mobile app, where they can create routines, manage tasks, and communicate in group chats.
The assistant is also aimed at caregivers looking after aging parents, offering a way to stay informed about their well-being even when living apart. Panasonic has collaborated with AARP to enhance Umi’s ability to support older adults. Additionally, the platform will connect users with wellness experts and integrate with partners such as Calm, Blue Apron, SleepScore Labs, and Precision Nutrition to help families build healthy habits.
Umi is expected to launch in the United States in 2025, with Panasonic positioning it as part of a broader wellness initiative. The partnership with Anthropic extends beyond consumer products, as Panasonic plans to integrate the Claude AI model into its own operations to enhance customer service, marketing, and coding efficiency.
Israeli startup Fermata, founded in 2020 by bioinformatics expert Valeria Kogan, is using AI and computer vision to monitor greenhouse crops for diseases and pests. The company’s software works with standard cameras, capturing images of plants twice a day and alerting farmers to potential infestations via an app. Initially considering robotic solutions, Kogan shifted focus after consulting with farmers, realising that simpler camera-based monitoring was more effective.
Based in Israel, Fermata has gained traction by prioritising farmer needs and keeping its AI training in-house, improving model accuracy. Partnering with major agricultural firms like Bayer and Syngenta, the company has deployed over 100 cameras and continues to expand. The startup recently secured a $10 million Series A investment from Raw Ventures, its existing investor, to scale operations and work towards profitability by 2026.
Plans for growth include strengthening the sales team and expanding beyond greenhouse tomatoes into new crops. Despite AI’s previous struggles in agriculture, Fermata’s practical approach and farmer-centric model have helped it carve a niche in the industry.
AMD has announced a range of new processors and graphics cards at CES 2025, including high-performance desktop CPUs, energy-efficient laptop chips, and AI-powered processors for next-generation Copilot+ PCs. The company’s latest flagship, the Ryzen 9 9950X3D, targets gamers and creators with 16 cores and speeds of up to 5.7GHz, offering an 8% performance boost in select games compared to its predecessor. AMD also introduced the Fire Range series for laptops and the Ryzen AI 300 and Ryzen AI Max chips, which integrate neural processing units for AI workloads.
The growing market for handheld gaming PCs has led to the release of AMD’s Ryzen Z2 series, optimised for portable devices. Meanwhile, the company’s new Radeon RX 9070 XT and RX 9070 GPUs, built on RDNA 4 architecture, promise improved ray tracing, AI acceleration, and better media encoding. AMD’s FidelityFX Super Resolution 4.0, designed to enhance gaming visuals with minimal latency, was also unveiled.
Expanding beyond hardware, AMD’s Adrenalin software now includes AI-powered features, such as image generation and local AI models for summarising documents. With a strong market presence and increasing demand for AI and gaming solutions, AMD’s 2025 lineup reflects its strategy to remain competitive across multiple segments. Ryzen AI and Fire Range: AMD’s big CES 2025 reveals
The growing use of AI in drug development is dividing opinions among researchers and industry experts. Some believe AI can significantly reduce the time and cost of bringing new medicines to market, while others argue that it has yet to solve the high failure rates seen in clinical trials.
AI-driven tools have already helped identify potential drug candidates more quickly, with some companies reducing the preclinical testing period from several years to just 30 months. However, experts point out that these early successes don’t always translate to breakthroughs in human trials, where most drug failures occur.
Unlike fields such as image recognition, AI in pharmaceuticals faces unique challenges due to limited high-quality data. Experts say AI’s impact could improve if it focuses on understanding why drugs fail in trials, such as problems with dosage, safety, and efficacy. They also recommend new trial designs that incorporate AI to better predict which drugs will succeed in later stages.
While AI won’t revolutionise drug development overnight, researchers agree it can help tackle persistent problems and streamline the process. But achieving lasting results will require better collaboration between AI specialists and drug developers to avoid repeating past mistakes.
Google has announced a major update to its TV operating system at CES 2025, integrating its Gemini AI assistant to deliver personalised news summaries. The new ‘News Brief’ feature will scrape news articles and YouTube headlines from trusted sources to generate a concise recap of daily events. Google plans to roll out the feature to both new and existing Google TV devices by late 2025.
The move marks Google’s deeper foray into AI-generated news, a space that has faced legal challenges from media companies over copyright concerns. While rival firms like OpenAI and Microsoft have been sued over unlicensed content use, Google’s News Brief does not currently display its sources, apart from related YouTube videos. AI-generated news has also faced accuracy issues, with previous AI models producing misleading or entirely false headlines.
Beyond news summaries, Google aims to make TVs more interactive, with Gemini allowing users to search for films, shows, and YouTube videos using natural language. Future Google TVs will include sensors to detect when users enter the room, enabling a more personalised experience. As the company continues expanding AI features in consumer technology, the success of News Brief may depend on how well it addresses content accuracy and transparency concerns.
The FBI has raised alarms about the growing use of artificial intelligence in scams, particularly through deepfake technology. These AI-generated videos and audio clips can convincingly imitate real people, allowing criminals to impersonate family members, executives, or even law enforcement officials. Victims are often tricked into transferring money or disclosing personal information.
Deepfake scams are becoming more prevalent in the US due to the increasing accessibility of generative AI tools. Criminals exploit these technologies to craft realistic phishing emails, fake social media profiles, and fraudulent investment opportunities. Some have gone as far as generating real-time video calls to enhance their deception.
To protect against these threats, experts recommend limiting the personal information shared online, enabling two-factor authentication, and verifying any unusual or urgent communications. The FBI stresses the importance of vigilance, especially as AI-driven scams become more sophisticated and harder to detect. By understanding these risks and adopting stronger security practices, individuals can safeguard themselves against the growing menace of deepfake fraud.
A notorious stretch of the A361 in Devon will receive £1 million in AI and speed camera technology to improve road safety. The investment, part of a £5 million grant from the Department for Transport (DfT), comes after the road was identified as ‘high risk,’ with three fatalities and 30 serious injuries recorded between 2018 and 2022. AI-powered cameras will detect offences such as drivers using mobile phones and failing to wear seatbelts, while speed cameras will be installed at key locations.
A pilot scheme last August recorded nearly 1,800 potential offences along the route, highlighting the need for stricter enforcement. The latest plans include three fixed speed cameras at Ilfracombe, Knowle, and Ashford, as well as two average speed camera systems covering longer stretches of the road. AI cameras will be rotated between different locations to monitor driver behaviour more effectively.
Councillor Stuart Hughes, Devon County Council’s cabinet member for highways, expressed pride in the region’s adoption of AI for road safety improvements. The remaining £4 million from the DfT grant will be allocated to upgrading junctions and improving access for pedestrians and cyclists along the A361.
