1. IRS Adopts AI to Enhance Taxpayer Services
The Internal Revenue Service (IRS) is actively deploying artificial intelligence (AI) agents to revolutionise its taxpayer services. This strategic move aims to improve efficiency and alleviate the workload on its existing staff, who are currently facing significant reductions. By integrating AI, the IRS expects to streamline processes, provide quicker responses to taxpayer queries, and ensure a smoother experience for everyone interacting with the agency. This initiative underscores a broader trend of government bodies leveraging advanced technology to optimise operations and enhance public service delivery amidst resource constraints.
Read full story
2. ChatGPT Atlas Gets Vertical Tabs & Google Search
ChatGPT Atlas has rolled out significant new features to enhance user experience. Key additions include vertical tabs, allowing for better organisation of multiple open tabs. Users will also find Google set as the default search engine, streamlining their search queries directly from the browser. Furthermore, the updated version supports seamless migration of extensions, making the transition for users easier. Integration with iCloud Keychain has also been implemented, providing secure and convenient password management for Apple users. These updates aim to improve productivity and convenience for its user base.
Read full story
3. AI Music Takes a Leap: Klay Vision Bags Major Record Deals
Klay Vision, a prominent AI music company, has officially secured licensing agreements with several major record labels. These landmark deals will enable Klay Vision to train its advanced Large Music Model (LMM) using a vast library of legally licensed music. This move addresses long-standing concerns in the music industry regarding copyright and fair use of existing musical works for AI development, paving the way for responsible innovation in algorithmic music creation. The collaboration is expected to profoundly impact how AI generates and integrates with the music ecosystem.
Read full story
4. Honor Launches AI Camera Robot Phone with Mechanical Gimbal
Honor has unveiled an innovative new smartphone, dubbed the ‘robot phone’, which integrates an artificial intelligence (AI) powered camera. This unique camera is mounted on a mechanical arm that seamlessly extends from the rear of the device, offering advanced photographic capabilities and potentially new modes of interaction. This design aims to enhance imaging stability and versatility, setting it apart in the competitive smartphone market.
Read full story
5. Google Launches Gemini 3.0, Claiming Superior LLM Performance
Google has officially released its latest large language model, Gemini 3.0. This new iteration is currently undergoing rigorous testing to validate its claims of significantly outperforming other leading Large Language Models (LLMs) available in the market. The tech giant aims to set a new benchmark in AI capabilities with Gemini 3.0, offering enhanced performance and advanced functionalities.
Read full story
6. AI’s Future: 900 Days to Decide, Says Stability AI CEO
Emad Mostaque, CEO of Stability AI, has issued a stark warning, stating that humanity has approximately 900 days to make critical decisions regarding the future of artificial intelligence. According to Mostaque, this timeframe is crucial before AI’s advancements reach a point where their impact on society becomes irreversible. His remarks highlight the urgent need for global discussions and regulatory frameworks to shape the ethical development and deployment of AI technologies. The implication is that without timely intervention, humanity risks losing control over the trajectory and consequences of artificial intelligence.
Read full story
7. AI Tricked by Poetic Prompts 62% of the Time
New research indicates that large language models (LLMs) are vulnerable to manipulation, with scientists successfully tricking them 62% of the time, merely by embedding harmful prompts within poetic structures. This finding highlights a significant security concern and a potential loophole in the current safeguarding mechanisms of AI, as attackers could exploit the subtle nature of poetry to bypass safety filters and elicit undesirable responses from the models. The study underscores the need for more robust and sophisticated detection methods to prevent such forms of adversarial attacks in the future.
Read full story
8. Gemini 3 Gets Stuck in 2024, Accuses Researcher of Trickery
Google’s advanced AI model, Gemini 3, recently displayed peculiar behaviour by refusing to acknowledge the current year as 2025. During an interaction, the AI model insisted it was still the year 2024 and even accused a researcher of attempting to trick it regarding the date. This incident highlights ongoing challenges and quirks in the development of sophisticated artificial intelligence systems, prompting further investigation into its temporal understanding and reasoning capabilities.
Read full story
9. NotebookLM Now Converts Notes to Slide Decks
NotebookLM has introduced a new feature allowing users to generate slide decks directly from their existing sources and notes. This innovative functionality is initially available to Pro users, enabling them to transform their research and compiled information into customizable presentations with ease. This update aims to streamline the process of creating professional-looking slides, making it simpler for users to organise and present their findings.
Read full story
10. Memo Robot Learns Household Chores from Real Families
The innovative Memo robot has been developed to assist with household tasks by learning from real-world scenarios. Unlike traditional lab-based training, Memo’s artificial intelligence is trained on an extensive dataset comprising 10 million daily routines observed from 500 diverse families. This unique approach allows Memo to understand and adapt to various household needs and preferences, making it highly effective for practical home assistance. Its ability to learn directly from human behaviour ensures it can perform tasks intuitively and efficiently, integrating seamlessly into everyday family life.
Read full story
11. Ubisoft Tests AI Characters for Voice-Controlled Gaming
Ubisoft is currently experimenting with an innovative AI technology that allows players to interact with non-player characters (APCs) using voice commands. This cutting-edge gaming experiment aims to revolutionise gameplay by enabling players to influence and personalise storylines through natural speech, offering a more immersive and dynamic experience. Early trials are underway to assess the potential of this voice-responsive AI in shaping the future of interactive entertainment and player agency within games.
Read full story
12. Google Integrates Ads into AI Search Results
Google has begun displaying sponsored results within its AI Mode responses, marking a significant step towards normalising advertisements in AI-powered search queries. This move aims to monetise the new AI search experience, just like traditional search results. Users may now see relevant ads appearing alongside AI-generated answers, potentially changing how information is consumed through Google’s AI.
Read full story
13. Amazon Unleashes Internal AI: Agent SOPs Now Open Source
Amazon Web Services (AWS) has announced the open-sourcing of its Agent Standard Operating Procedures (SOPs). These SOPs were previously used internally by Amazon to streamline operations, democratise AI expertise across various teams, and automate repetitive tasks. By making Agent SOPs publicly available, AWS aims to empower a wider range of developers and businesses to leverage AI’s potential, simplifying complex workflows through natural language interactions. This move is expected to foster innovation and efficiency in AI agent development globally.
Read full story
14. Louvre Heist Reveals AI Weaknesses
A recent robbery at the iconic Louvre Museum has brought to light how thieves exploited common biases in pattern recognition, affecting both humans and artificial intelligence systems. This incident underscores critical vulnerabilities in current AI security protocols, demonstrating that sophisticated criminals can leverage predictable human and algorithmic responses to their advantage. Experts suggest that such biases, often overlooked, provide avenues for illicit activities by manipulating expected behaviours.
Read full story
15. Google CEO: AI Agents Set to Revolutionise Tech in a Year
Google’s CEO, Sundar Pichai, has made a significant prediction, anticipating the emergence of highly sophisticated AI agents within the next year. These advanced artificial intelligence systems are expected to bring about a transformative impact on various technological sectors. This projection coincides with Google’s substantial investment of $90 billion in infrastructure, highlighting the company’s commitment to developing and supporting cutting-edge AI technologies and capabilities.
Read full story
16. Spotting AI Music: Key Indicators Revealed
Music generated by artificial intelligence (AI) exhibits specific characteristics in its composition and execution. These distinct markers allow platforms to identify and flag such content. Experts are working to refine detection methods as AI music generation becomes more sophisticated. This helps ensure transparency and proper accreditation in the digital music landscape.
Read full story
17. Caltech Unveils Drone-Launching Humanoid Robot
Researchers at Caltech have developed an innovative humanoid robot that comes equipped with a drone backpack. This advanced robot is designed for versatile operations, featuring capabilities for both driving on the ground and launching an integrated drone for aerial tasks. This development marks a significant step in robotics, combining terrestrial and aerial mobility in a single unit.
Read full story
18. AI Uncovers New Lion Roar Type
Scientists have leveraged artificial intelligence (AI) to discover a previously unrecognised “intermediate” lion roar. This breakthrough came after researchers analysed over 3,000 lion vocalisation recordings. The AI system successfully identified this unique roar type with an impressive 95% accuracy rate, providing new insights into lion communication patterns and behaviour. This development could help in better understanding and conserving lion populations worldwide.
Read full story
