Meta is significantly ramping up its artificial intelligence efforts by forming a new applied AI engineering organization. This group will work closely with Meta Superintelligence Labs, aiming to accelerate the pursuit of superintelligence and faster AI model development. Maher Saba, a vice president at Reality Labs, leads this new unit, reporting to CTO Andrew Bosworth. Saba emphasizes that building robust AI models requires real-world data and feedback, not solely researchers and computing power, with the team focused on creating the necessary data engine.
Beyond Meta's internal advancements, the broader AI landscape sees diverse applications and challenges. Current AI and Bhashini, for instance, unveiled an open-source, handheld AI device at the India AI Impact Summit. This device can identify objects and respond in multiple languages offline, with CEO Ayah Bdeir highlighting its role in offering an alternative to Big Tech-dominated AI and addressing data privacy concerns. Meanwhile, Burger King is piloting an AI chatbot named Patty, powered by OpenAI, in about 500 restaurants to coach employees on tasks and customer interactions, tracking 'friendliness' without recording conversations.
Academic institutions are also actively engaging with AI's implications. Washington University in St. Louis launched its +AI initiative, led by Provost Mark West, to integrate AI into research and education while upholding core values. This includes providing access to tools like Gemini for Education. Elon University hosted an 'AI & The Body' panel on March 3, where professors discussed AI as a productivity tool in fields like health, comparing concerns about AI making people less intelligent to the advent of calculators. Locally, the Left Hand Valley Courier is launching a series to explore AI's impact on journalism, schools, businesses, and personal lives within its community.
However, the rapid adoption of AI brings significant security concerns. Federal cybersecurity leaders are addressing new risks like data poisoning and prompt injection, with a red-teaming exercise showing 90% of AI systems compromised in under 90 minutes. Experts stress the need for stronger governance, including Model Bills of Materials (MBOMs) and continuous red-teaming. Attackers are increasingly targeting the infrastructure surrounding AI systems, rather than just the models, exploiting vulnerabilities in AI 'skills' and training data. This comes as 83% of organizations plan to deploy agentic AI, but only 29% feel prepared. Even hobbies are affected, as jigsaw puzzle enthusiasts find AI-generated art diminishes their experience due to a lack of detail and coherence.
Key Takeaways
- Meta is forming a new applied AI engineering organization, led by Maher Saba, a vice president at Reality Labs and reporting to CTO Andrew Bosworth, to accelerate superintelligence development and improve AI models with real-world data.
- Current AI and Bhashini unveiled an open-source, handheld AI device, championed by CEO Ayah Bdeir, offering an offline, multi-language alternative to Big Tech AI with designs available on GitHub.
- Burger King is piloting an OpenAI-powered AI chatbot named Patty in approximately 500 restaurants to coach employees on tasks and customer interactions, tracking 'friendliness.'
- Washington University in St. Louis launched its +AI initiative, led by Provost Mark West, to integrate AI into research and education, including access to tools like Gemini for Education.
- Federal cybersecurity leaders are addressing significant AI security risks, such as data poisoning and prompt injection, after a red-teaming exercise compromised 90% of AI systems in under 90 minutes.
- Organizations face a readiness gap for AI security, with 83% planning agentic AI deployment but only 29% feeling prepared for attacks targeting AI infrastructure and training data.
- Elon University hosted a panel discussing AI's role as a productivity tool in professional fields like health and research, emphasizing human expertise remains vital.
- Jigsaw puzzle enthusiasts are encountering issues with AI-generated art due to its lack of detail and coherence, which detracts from the hobby's tactile experience.
- The Left Hand Valley Courier is launching a series to explore AI's impact on local journalism, schools, businesses, and personal lives.
Meta Forms New AI Team for Superintelligence Push
Meta is creating a new applied AI engineering organization to speed up its pursuit of superintelligence. This new group will work closely with Meta Superintelligence Labs, which focuses on developing advanced AI models. The organization will have a flat structure with up to 50 employees per manager, reporting to CTO Andrew Bosworth. Maher Saba, vice president at Reality Labs, will lead the team, which aims to improve AI models using real-world data and feedback. The team will build the data engine needed to make AI models better and faster.
Meta Creates New AI Unit to Speed Up Model Development
Meta is establishing a new applied artificial intelligence engineering group to work with its Superintelligence Lab. This team aims to make AI models develop faster. It will include a team for building interfaces and tools, and another for tasks like generating data and evaluating models. Maher Saba, a vice president in Meta's Reality Labs, will lead the new unit. Saba stated that building great models requires real-world data and feedback, not just researchers and computing power. This initiative seeks to improve AI models through reinforcement learning and post-training efforts.
Open Source AI Hardware Challenges Big Tech's Dominance
Current AI and Bhashini have unveiled a new open-source, handheld AI device that can identify objects and respond in multiple languages, even offline. This device, demonstrated at the India AI Impact Summit, aims to provide an alternative to AI systems dominated by Big Tech. The project, supported by governments and foundations, will release its designs on GitHub, empowering anyone to build their own AI applications. CEO Ayah Bdeir highlighted the need for open, customizable AI platforms to counter data privacy concerns with Big Tech devices. The initiative seeks to make AI development more equitable and scalable for diverse communities.
Elon University Panel Explores AI in Professional Fields
Elon University hosted an 'AI & The Body' panel on March 3, discussing artificial intelligence's role in professional industries like research and health. Professors Shannon Duvall, Bill Evans, and Antoinette Polito shared insights, emphasizing AI as a tool for increasing productivity and access. They addressed common concerns about AI making people less intelligent, comparing it to the advent of calculators. The panelists also noted AI's potential to assist in fields like radiology and pathology, while highlighting the importance of human expertise. The discussion is part of a three-part series exploring AI's impact.
Burger King Tests AI Chatbot Patty for Employee Coaching
Burger King is piloting an AI chatbot named Patty, part of its internal BK Assistant platform, to help employees with tasks and evaluate customer interactions. Patty, powered by OpenAI, can answer questions on food prep and maintenance, and update digital menus if equipment fails. The AI also tracks 'friendliness' by listening for key phrases like 'welcome' and 'thank you.' Burger King states Patty is a coaching tool to support staff, not to record conversations or penalize individuals. The pilot program involves about 500 restaurants, aiming to improve overall service patterns and hospitality.
AI Art Creates Problems for Jigsaw Puzzle Fans
Jigsaw puzzle enthusiasts are finding that AI-generated art is diminishing their hobby due to a lack of detail and coherence. Puzzlers meticulously examine images for specific details, a process that highlights the flaws in AI art, such as incorrect features or misplaced elements. Companies are using AI art for puzzles because it's cheaper than hiring human artists, leading to more AI-generated puzzles on the market. Many hobbyists view puzzling as a tactile, analog escape from the digital world, and AI art undermines this experience. The AI art, while sometimes visually appealing overall, often lacks the fine detail and logical consistency that makes puzzles satisfying to solve.
Local Newspaper Series Explores AI's Community Impact
The Left Hand Valley Courier is launching a series examining how artificial intelligence affects their local community. The series will cover AI's influence on journalism, schools, businesses, and personal lives. While the publication uses traditional methods, it acknowledges the evolving nature and value of AI technology. The articles will critically assess AI's current impact and its ongoing development within the community. This initiative aims to inform readers about the significant changes AI is bringing to their daily lives and local institutions.
Washington University Launches +AI Initiative
Washington University in St. Louis has launched a new universitywide initiative called +AI, led by Provost Mark West. The program aims to balance the potential of artificial intelligence with the university's core values of creativity and academic expertise. +AI will explore how AI can enhance research and education without replacing human talent. The initiative includes new learning modules, access to AI tools like Gemini for Education, and faculty insight groups to guide future planning. A university-wide event is also planned for the week of March 23 to discuss AI's role.
Government Leaders Address AI Security Risks and Governance
Federal cybersecurity leaders are grappling with new AI security risks, including data poisoning and prompt injection, as AI adoption grows in government agencies. A panel at Zscaler's Public Sector Summit highlighted that many AI systems are vulnerable, with a recent red-teaming exercise showing 90% compromised in under 90 minutes. Experts emphasized the need for stronger governance, suggesting solutions like threat modeling, incorporating governance into procurement, and using Model Bills of Materials (MBOMs). The discussion also stressed the importance of continuous AI red-teaming and educating the workforce on AI's capabilities and risks.
AI Systems Face New Security Threats in 2026
Attackers are increasingly targeting the infrastructure surrounding AI systems, rather than the models themselves, posing new security risks. These attacks focus on how AI interacts with tools, data, and codebases, potentially leading to data leaks or unintended actions. A significant gap exists in organizations' readiness to securely operate AI, with 83% planning to deploy agentic AI but only 29% feeling prepared. Vulnerabilities are found in AI 'skills,' training data, and protocols connecting models to tools. Additionally, attackers are using AI to aid in cyber operations like creating phishing messages and malicious code.
Sources
- Meta advances AI goals with new applied engineering team
- Meta Creates New AI Unit to Accelerate Model Development
- Open-source AI hardware could weaken Big Tech’s grip on AI
- AI & The Body: A discussion on artificial intelligence in professional industries
- Burger King To Use AI Chatbot to Track Employee Friendliness
- The hobby that AI is ruining for its fans
- Local Impact of Artificial Intelligence
- University introduces +AI academic initiative
- Federal Leaders Confront the Next Wave of AI Security Risks
- Security in 2026: New Ways Attackers Are Exploiting AI Systems - The National CIO Review
Comments
Please log in to post a comment.