Google has launched Gemma 4, a new set of open AI models available under an Apache 2.0 open-source license. These models, built on Gemini 3 technology, are designed to run on a wide range of devices, from smartphones to powerful workstations. Gemma 4 comes in four sizes, including E2B and E4B variants optimized for mobile and edge devices, and larger versions like 26B MoE and 31B Dense that perform well on industry benchmarks for complex tasks such as reasoning, code generation, and processing various data types like images, video, and audio.
The popularity of Google's AI Edge Gallery app, which features Gemma 4, highlights a growing interest in on-device AI, reaching the top 10 on the App Store's productivity charts. This app allows for fast, offline performance with minimal memory usage, supporting over 140 languages for AI chats, image analysis, and audio transcription. Meanwhile, Tredence and Google Cloud are expanding their partnership to help businesses implement enterprise-grade AI solutions at scale, leveraging tools like Gemini Enterprise and Vertex AI to operationalize AI and drive measurable business outcomes.
The rapid expansion of AI technology brings both opportunities and challenges across various sectors. Oracle's co-chief executive, Mike Sicilia, emphasizes that avoiding the AI race is the company's biggest risk, signaling a strong commitment to AI investments. However, the AI boom also impacts energy infrastructure, with MIT professor Sinan Aral discussing the need for hybrid models to manage grid power and the global buildout of data centers.
Concerns about AI's effect on employment are also prominent. US tech companies are cutting jobs while simultaneously increasing AI investments, a trend that experts believe could reshape work, though AI is not yet replacing large numbers of workers. A report indicates that Minnesota workers face high exposure to generative AI, with about 17% of the state's workforce, or 500,000 people, potentially seeing their jobs altered or replaced. This has prompted lawmakers to prioritize AI regulation to protect workers, as disruptions are anticipated.
AI also presents a double-edged sword for smartphone security, with sophisticated AI-generated phishing attacks on the rise, affecting 27% of consumers in the past year. Conversely, AI is being used for on-device protections like natural language checks and deepfake detectors, with Google's Scam Detection identifying malicious messages. In the public sector, the focus for AI by 2026 will be on data interoperability, outcome-based oversight, and secure AI enclaves, ensuring AI is scalable, human-controlled, and compliant. Furthermore, a study warns of 'algorithmic greenwashing,' where AI tools trained on corporate sustainability reports can unintentionally spread vague environmental language, highlighting the need for explicit constraints on AI outputs.
Key Takeaways
- Google launched Gemma 4, a new set of open AI models under an Apache 2.0 license, built on Gemini 3 technology, designed for diverse devices from smartphones to workstations.
- Google's AI Edge Gallery app, featuring Gemma 4 variants like E2B and E4B for smartphones, has climbed into the App Store's top 10 productivity charts, indicating strong interest in on-device AI.
- Tredence and Google Cloud are expanding their partnership to help businesses operationalize enterprise-grade AI solutions at scale, utilizing Google Cloud's Gemini Enterprise and Vertex AI.
- Oracle's co-chief executive, Mike Sicilia, states that avoiding the ongoing AI race is the company's biggest risk, committing to significant AI investments.
- US tech companies are cutting jobs while increasing AI investments, a trend that experts believe will reshape work, with about 17% of Minnesota's workforce (500,000 people) facing potential job alterations or replacement by AI.
- AI presents a dual challenge for smartphone security, with 27% of consumers reporting AI-generated phishing scams, while AI is also used for on-device defenses like Google's Scam Detection.
- MIT professor Sinan Aral highlights the impact of the AI boom on grid power, suggesting a hybrid model for energy management amidst global data center expansion.
- Public sector organizations will focus on data interoperability, outcome-based oversight, and secure AI enclaves by 2026 to ensure scalable, human-controlled, and compliant AI deployments.
- A study warns of 'algorithmic greenwashing,' where AI tools trained on corporate sustainability reports can unintentionally spread vague or misleading environmental language, requiring explicit constraints on AI outputs.
Google releases Gemma 4 AI models for all devices
Google has launched Gemma 4, a new set of open AI models that can run on various devices, from smartphones to workstations. These models are available under an Apache 2.0 open-source license and are designed for advanced tasks like reasoning and processing different types of data. Gemma 4 builds on the technology used in Gemini 3 and comes in four sizes, with the larger models performing well on industry leaderboards. The smaller versions are optimized for mobile and edge devices, while the larger ones work on powerful GPUs. These models can handle complex logic, generate code, and process images, video, and audio, with extended context windows for long content.
Google AI app climbs App Store charts with on-device Gemma 4
Google's AI Edge Gallery app has reached the top 10 on the App Store's productivity charts, showing growing interest in running AI directly on devices. The app features the Gemma 4 model, including variants like E2B and E4B designed for smartphones. These models offer fast, offline performance using minimal memory and can process text, images, and audio in over 140 languages. Users can engage in AI chats, ask questions about images, transcribe audio, and use agent skills for on-device workflows. The app's popularity highlights a trend towards on-device AI due to privacy, offline needs, and faster processing.
Google's Gemma 4 AI models now available for diverse hardware
Google has introduced Gemma 4, a new collection of open AI models designed for use on various hardware, including mobile phones and personal computers. These models are released under an Apache 2.0 open-source license, supporting advanced reasoning and multimodal data processing. Gemma 4 is built using the same technology as Gemini 3 and comes in four sizes: E2B, E4B, 26B MoE, and 31B Dense. The models excel in complex tasks and have shown strong performance on industry benchmarks. Smaller versions are optimized for efficiency on mobile and edge devices, while larger versions run on high-end GPUs, with some compatible with consumer graphics cards.
Author faces 'professional identity purgatory' after AI job loss
After a 30-year career ended due to an acquisition, author Geoff Curtis describes entering 'professional identity purgatory,' a state of transition without a clear direction. He argues that AI is pushing millions into similar situations, challenging not just jobs but the fundamental sense of professional relevance. Curtis advises against filling the void with constant activity, suggesting instead to embrace a provisional identity and redefine expertise beyond technical skills. He emphasizes that judgment, relationships, and context remain uniquely human and valuable, even as AI automates tasks.
Oracle CEO: Avoiding AI race is biggest risk
Mike Sicilia, co-chief executive of Oracle, stated that the company's greatest risk would be to ignore the ongoing AI race. He expressed full commitment to Oracle's investment plans in embracing new AI technologies. This indicates a strong strategic focus on integrating artificial intelligence into Oracle's future operations and offerings.
AI is a double-edged sword for smartphone security
Artificial intelligence presents both risks and solutions for smartphone security. AI-generated phishing attacks are becoming more sophisticated, with 27% of consumers reporting such scams in the past year, particularly in English-speaking countries. However, AI is also being used to develop on-device protections like natural language checks and deepfake detectors. Google's Scam Detection uses AI to identify malicious messages and calls, though consumer acceptance of AI features in smartphones remains low in many regions. The development of AI in security is a race between attackers using AI tools and defenders implementing AI-based defenses.
MIT professor discusses hybrid AI and grid power
Sinan Aral, a professor at MIT and director of the Initiative on the Digital Economy, discussed the impact of the AI boom on grid power. He suggested that a hybrid model offers the best of both worlds for managing energy needs. Aral also touched upon the buildout of data centers globally and the potential effects of geopolitical events on regional data centers. His insights highlight the complex relationship between rapidly growing AI technology and energy infrastructure.
Tech job cuts rise as companies bet on AI
US tech companies are cutting jobs while increasing investments in artificial intelligence, a move experts suggest is part of a large-scale experiment that could reshape work. Despite AI's ability to speed up tasks like coding and data analysis, many experts believe AI is not yet capable of replacing large numbers of workers. The trend involves increased use of AI tools by employees, sometimes leading to more code to review and potential job insecurity. This shift may set a precedent for job cuts across various industries as companies prioritize AI investments and efficiency gains.
Minnesota workers face high AI job risk, report finds
A new report indicates that Minnesota workers have the highest exposure to generative AI in the Midwest and rank tenth nationally. Approximately 17% of the state's workforce, around 500,000 people, could have their jobs significantly altered or replaced by AI. Lawmakers are prioritizing AI regulation to protect workers from potential job displacement. While AI is not yet causing widespread layoffs, experts anticipate disruptions are inevitable. The University of Minnesota is working to prepare students for the evolving job market, emphasizing AI literacy and adapting education to different fields.
Tredence and Google Cloud boost AI partnership
Tredence and Google Cloud are expanding their global partnership to help businesses implement enterprise-grade AI solutions at scale. This collaboration focuses on operationalizing AI, moving beyond experimentation to deliver practical applications that enhance agility and growth. They will leverage Google Cloud's AI tools, including Gemini Enterprise and Vertex AI, along with Tredence's AI accelerators and expertise. The goal is to transform enterprise data into strategic assets and enable organizations to deploy self-reasoning AI systems effectively, driving measurable business outcomes across industries.
Public sector AI to focus on data, oversight, and security in 2026
Public sector organizations will focus on three key shifts for AI in 2026: data interoperability, outcome-based oversight, and secure AI enclaves. With tight budgets and increasing oversight, agencies must ensure their data is AI-ready and accessible. Oversight will shift towards real-time transparency of AI model usage and impact, prioritizing projects based on return on investment. Furthermore, AI deployment will move towards secure, domain-specific enclaves for critical applications, emphasizing scalability and human control to ensure compliance and avoid vendor lock-in.
AI can cause 'algorithmic greenwashing,' study finds
A new study warns of 'algorithmic greenwashing,' where AI tools trained on corporate sustainability reports unintentionally spread vague or misleading environmental language. Researchers found that an AI agent built to help companies with nature and biodiversity efforts defaulted to generating reassuring but unhelpful responses, mimicking common corporate sustainability jargon. This bias towards 'constructive optimism' means AI might reproduce greenwashing language instead of providing concrete guidance. The study suggests that AI outputs need explicit constraints to ensure they offer useful, non-greenwashing advice.
Sources
- Google unveils Gemma 4 open AI models for diverse hardware use
- Google AI Edge Gallery Breaks Into Top 10 On App Store As Users Try Out On-Device Gemma 4 Model
- Google unveils Gemma 4 open AI models for diverse hardware use
- AI is about to send millions to ‘professional identity purgatory.’ Here’s what I discovered after my 30 year career crashed to a halt
- ‘The biggest risk for us at Oracle would be avoiding the AI race’
- AI is both a friend and foe to the future of smartphone security
- MIT's Sinan Aral on grid power & the AI boom: A hybrid model gets the best of both worlds
- Tech companies are cutting jobs and betting on AI. The payoff is far from guaranteed
- Report says Minnesota workers face highest generative AI exposure in the Midwest
- Tredence Expands Global Strategic AI Partnership with Google Cloud to Accelerate Enterprise-Grade AI Adoption
- Public Sector AI: 3 Shifts for 2026
- Algorithmic greenwashing: Lessons from building an AI agent for nature
Comments
Please log in to post a comment.