Tech News Digest: March 4, 2026

‘Merlin’ AI Revolutionizes Medical Diagnostics with Automated CT Scan Analysis

A research team funded by the National Institutes of Health (NIH) has developed a versatile machine learning model named Merlin that analyzes 3D abdominal computed tomography (CT) scans for a wide range of diagnostic tasks. This powerful AI in healthcare is capable of identifying anatomical features and, more significantly, predicting disease onset years in advance. Merlin was trained on a large dataset of patient CT scans linked with radiology reports and medical diagnosis codes from Stanford University. In testing, the general-purpose model surpassed other automated tools specifically designed for the tasks it was evaluated on, signaling a major step forward in streamlining clinical workflows and enabling earlier disease diagnosis.

Sources:

Mobilicom Highlights Advanced Drone & Robotics Cybersecurity Solutions

Mobilicom, a leading provider of cybersecurity solutions for drones and robotics, was a Gold Sponsor at the Loitering Munition Systems Summit in Huntsville, Alabama. The company showcased its end-to-end cybersecure solutions for unmanned systems. At the summit, Mobilicom’s US Technical Sales Manager delivered a key presentation on maintaining secured communications for loitering munition missions in contested electronic warfare environments. The event gathered experts to discuss the latest advancements in loitering munitions systems for defense, where Mobilicom emphasized the critical importance of secure and resilient communications for the mission success of autonomous and swarm-based capabilities.

Sources:

OpenAI Releases GPT-5.3 Instant to Enhance ChatGPT’s Accuracy and Tone

OpenAI has rolled out a significant update for its popular chatbot, introducing GPT-5.3 Instant as the new default model for ChatGPT. This update aims to address user feedback regarding the large language model’s (LLM) conversational style, reducing what some have described as a ‘cringe’ tone. The new version is engineered to be more direct, with fewer unnecessary disclaimers and a more natural conversational flow. Key improvements include a sharp reduction in hallucinations, with OpenAI reporting a 22.5% decrease when using web search and a 9.6% decrease without it. GPT-5.3 Instant also demonstrates superior contextualization of web information, balancing online search results with its internal knowledge base. The update is now live for all ChatGPT users.

Sources:

COSO Publishes New Framework for Generative AI Governance and Internal Control

The Committee of Sponsoring Organizations of the Treadway Commission (COSO) has published new guidance on governing generative artificial intelligence. Titled ‘Achieving Effective Internal Control Over Generative AI (GenAI),’ the roadmap adapts COSO’s existing Internal Control-Integrated Framework to address the unique risks posed by GenAI. This guidance provides a structured approach for organizations to ensure their use of generative AI is responsible and supports reliable operations, reporting, and compliance. It emphasizes that while the technology is new, the fundamental principles of internal control and AI risk management still apply. The framework is designed to be practical and audit-ready, helping organizations navigate the rapidly evolving landscape of AI governance.

Sources:

EU Launches €75M EURO-3C Project to Build Federated Telco Edge Cloud

The European Commission has introduced the EURO-3C project at Mobile World Congress 2026, a €75 million initiative funded by the Horizon Europe research program. The project aims to create Europe’s first large-scale federated Telco Edge Cloud infrastructure, which will integrate telecommunications networks, edge computing, and cloud infrastructure into a unified platform. This initiative is designed to bring computing power closer to end-users and reduce the European Union’s dependency on third-country providers, bolstering Europe’s digital sovereignty. The project involves 87 consortium members, including major telecom operators, cloud service providers, and equipment manufacturers, aligning with the goals of the proposed Digital Networks Act.

Sources:

Breakthrough Method Simplifies Analysis of Quantum Computer Processes

An international research team has developed a new method called compressive quantum process tomography (CQPT) to more efficiently analyze the processes inside quantum computers. This technique is crucial for verifying the correct functioning of quantum gates and circuits, identifying hardware errors, and supporting calibration and quantum error correction. Traditional methods for this analysis, known as quantum process tomography (QPT), are costly and become impractical as quantum systems scale. The new CQPT framework offers a more scalable alternative, with the potential to make quantum tomography more efficient for larger systems. The findings were published in Advanced Quantum Technologies.

FIU Develops Quantum-Safe Encryption to Defend Against Future Cyber Threats

Researchers at Florida International University (FIU) have created a new quantum-safe encryption system to protect digital content from the threat of future cyberattacks by quantum computers. This post-quantum cryptography (PQC) development responds to concerns that powerful quantum computers could break current encryption standards, compromising vast amounts of sensitive data. The system combines quantum encryption with secure internet transmission to protect against both current and future hacking methods. In tests, this new method showed a 10–15% improvement over other advanced encryption techniques by significantly reducing data patterns that hackers could potentially exploit.

Sources:

Huawei Open-Sources A2A-T Protocol to Standardize AI Agent Communication in Telecom

At MWC 2026, Huawei announced it will open-source the core software for the Agent-to-Agent for Telecom (A2A-T) protocol. This initiative is designed to accelerate the adoption of a standardized communication method for AI agents within the telecommunications industry. The A2A-T protocol, co-authored by global telecom partners, aims to create a unified interaction framework for multi-agent collaboration. By open-sourcing the protocol, Huawei intends to move the standard from a theoretical specification to a practical, deployable solution. The project will include a Protocol SDK, a Registry Center for authentication, and an Orchestration Center with low-code tools, addressing key challenges in automated operations and secure agent interactions.

Sources: