NerdNewsFebruary 06, 2025 |
News & Trends
OpenAI Partners with National Labs
OpenAI has partnered with the US National Laboratories to deploy its advanced AI model on a supercomputer, aiming to supercharge scientific breakthroughs in areas like materials science, renewable energy, and national security. The partnership will provide 15,000 scientists with access to OpenAI's latest reasoning models, with a focus on accelerating US global tech leadership and stabilizing the power grid.
The biggest breach of US government data is under way
Elon Musk's team, DOGE, has gained access to sensitive US government data, including personal info of millions of Americans. The team's security clearances and cybersecurity practices are under question, raising concerns about data protection and national security.
Amazon reportedly gears up to release next-gen Alexa
Amazon is planning to release a next-gen, AI-powered Alexa experience that can respond to multiple commands in sequence. The new Alexa will be available for a limited number of users for free, with a potential $5 to $10 monthly fee. The current 'Classic Alexa' will still be available.
Google Lifts Self-Imposed Ban on Using AI for Weapons and Surveillance
Google has lifted its self-imposed ban on using AI for weapons and surveillance, marking a significant shift in the company's approach to AI development. The move comes after Google faced pressure over its involvement in Project Maven, a US Department of Defense contract that used AI for drone imaging. Google's updated AI principles now focus on 'bold innovation' and 'responsible development and deployment', but no longer explicitly rule out the use of AI for weapons or surveillance.
Meta Teaches AI Models to Prioritize
Meta AI and the University of Illinois Chicago propose a new technique to train models to allocate inference budgets based on query difficulty, resulting in faster responses and reduced costs. The technique uses sequential voting and adaptive sequential voting to improve model efficiency. A reinforcement learning algorithm, Inference Budget-Constrained Policy Optimization, is also introduced to optimize model responses within an inference budget constraint. |
Options & Tutorials
Container Queries Unleashed
Container queries allow for more flexible and dynamic layouts. They enable developers to style elements based on the size of their container, rather than the viewport. This article explores a useful pattern for using container queries to create responsive designs, and how to select which container to use with the 'container-name' property.
How Attackers Abuse S3 Bucket Namesquatting — And How to Stop Them
Attackers exploit predictable S3 bucket names to access or hijack buckets. This can lead to data breaches, traffic redirection, and admin account creation. To prevent this, customize S3 bucket names, ensure they are not public, and use tools like Varonis to automate security processes and detect misconfigurations.
Google’s Gemini app adds access to ‘thinking’ AI models
Google's Gemini app has been updated to version 2.0, which includes the 'Flash Thinking' AI model that can explain its reasoning and interact with other Google apps like YouTube. The update also brings cheaper access for developers and a new low-cost model called 2.0 Flash-Lite. Additionally, Google is releasing an experimental version of Gemini 2.0 Pro, which provides better factuality and stronger performance for coding and mathematics-related tasks.
Taking LLMs out of the Black Box
Ines Montani discusses using human-in-the-loop distillation to take large language models out of the black box, making them more transparent, efficient, and reliable. She provides practical tips and case studies on how to apply this technique to real-world applications.
InfoQ's New Certification for Senior Developers and Architects
InfoQ is introducing a hands-on software architecture certification at QCon London 2025, focusing on practical skills for senior developers and architects. The certification, led by Wesley Reisz, will combine conference sessions with a hands-on workshop on real-world architectural challenges. |
Launches & Tools
Apple Silicon M5 Chips
South Korean media claims Apple's M5 chips are in mass production. The M5 chip is expected to be used in the iPad Pro, Macs, and Apple Vision Pro products, with a possible release in late 2025. The chip will retain a similar architecture to the M1-M4 chips but may feature a new packaging process called System-in-Integrated-Chips-Molding-Horizontal (SoIC-mH).
Boston Dynamics Partners with Former CEO to Advance Atlas Robot
Boston Dynamics partners with the Robotics & AI Institute to improve the learning of its Atlas humanoid robot through reinforcement learning, aiming to enhance its ability to perform complex tasks like dynamic running and full-body manipulation of heavy objects.
CodeSignal Launches AI-Powered Soft Skills Guide
CodeSignal launches soft skill courses with AI-powered guide to practice professional skills and work-life scenarios. The platform offers over 100 courses, including leadership training and sales skills assessment, with personalized feedback from AI assistants.
iOS App Store apps with screenshot-reading malware found
Kaspersky discovered 'SparkCat' malware in multiple iOS apps that uses OCR tech to extract text from images, including crypto wallet passwords or recovery phrases, and sends them back to attackers.
Amazon Announces February Event
Amazon has sent out invites for a February event, which may mark the launch of its long-delayed smarter version of Alexa. The event, hosted by Panos Panay, will take place on February 26th. The company has been working on a supercharged Alexa digital assistant powered by generative AI, but its release has been delayed. The event invites spell out 'Alexa' when combined, hinting at the possible launch of the new assistant. |
Quick Links
Elon Musk's Secretive Government IT Takeover
Elon Musk and his team have gained access to critical US government IT systems, raising concerns about privacy and potential conflicts of interest. Musk's team, part of the newly created Department of Government Efficiency (DOGE), now has access to the Treasury Department, Small Business Administration, and Office of Personnel Management systems, containing sensitive personal information of millions of Americans. The move has sparked concerns about Musk's influence over the federal budget and potential for abuse of power.
Netgear fixes critical bugs as Five Eyes warn about break-ins at the edge
Netgear has fixed two critical bugs in its routers and access points, with the US, UK, Canada, Australia, and other countries warning about the risks of edge device break-ins. The vulnerabilities, with CVSS scores of 9.6 and 9.8, could allow authentication bypass and remote code execution. National security agencies have issued guidance on securing edge devices, emphasizing the importance of enhanced logging and forensic data gathering.
Workday Announces Restructuring and Job Cuts
Workday is cutting 1,750 jobs, about 8.5% of its workforce, as part of a restructuring effort. The company cites the need to adapt to a new environment and invest in areas like AI. Despite no surprises in its recent financial results, Workday will exit some owned office spaces and focus on strategic hiring. Affected employees will receive support, including a minimum of 12 weeks of pay and additional benefits.
Rackspace moving workloads off VMware to address Broadcom bills
Rackspace is moving some of its back-office workloads off VMware and onto Platform9's Private Cloud Director due to rising software bills under Broadcom's licensing changes. The move aims to run infrastructure more efficiently and reduce costs. Platform9's tool, vJailbreak, automates migrations away from VMware and has already been adopted by a Fortune 500 company to migrate over 40,000 virtual machines.
XOi raises $230M, acquires Specifx
XOi, a company providing software for field service technicians, has raised $230 million from KKR and acquired Specifx to expand its tech and datasets, aiming to improve maintenance and repair services for its users. |
Share NerdNewsShare your affiliate link to get commission!
Thanks for reading,
If you dont want to receive future editions of NerdNews, |