The Sydney AI Safety Fellowship 2026 has commenced a 10-week hybrid program in Sydney, aimed at addressing the risks of transitioning to advanced AI systems. Running from January 10 to March 1, the fellowship focuses on critical issues such as loss of control, AI-enabled pandemics, and societal-scale cyberattacks rather than general AI literacy. This initiative runs parallel to the Australian Government's launch of the Australian Artificial Intelligence Safety Institute (AISI), marking a shift toward specialized, risk-focused training within the local AI sector.

As AI development moved from general chatbots to actionable agents in 2025, the vocabulary of the industry shifted. This report explains ten critical terms—from reasoning models and small language models (SLMs) to content provenance—that defined the year’s technological and regulatory landscape.

Industry reporting for 2025 identifies Application Programming Interfaces (APIs) as the critical infrastructure connecting AI agents to real-world systems. Data from the Postman State of the API Report suggests that while AI adoption is widespread, API design and security measures often lag behind, raising significant concerns about unauthorized access and automated tool usage.

The Higher Education Research and Development Society of Australasia (HERDSA) has circulated a notice regarding an online workshop on digital ethics and Generative AI hosted by Hong Kong Baptist University. Scheduled for 14 January 2026, the session will address student perspectives on AI usage in mental health support and assessment, referencing frameworks for responsible AI in education.

As generative AI shifts from brainstorming to building, Figma and Manus present diverging paths for product development. This report examines Figma’s integration of the Model Context Protocol (MCP) to bridge designs and code, versus Manus AI’s all-in-one agent workflow powered by Google’s Nano Banana Pro. Beyond features, the analysis addresses critical governance concerns, including Figma’s late-2025 data litigation and the regulatory scrutiny surrounding Manus’s corporate structure.

The University of Melbourne is launching a new initiative, Cross-disciplinary AI Readiness for Health, Human Services and Teacher Education (CAIR), to embed generative AI capability into professional training. Led by Dr. Jessica Boyce and funded by a University Teaching and Learning Initiative grant, the project involves a collaboration between the Faculty of Medicine, Dentistry and Health Sciences and the Faculty of Education. CAIR aims to deliver a digital learning resource across more than 10 programs by early 2027, focusing on the ethical and strategic application of AI in people-facing professions.

Google’s "Double check response" feature provides a secondary verification layer for AI-generated content. By cross-referencing claims against live Google Search results, the tool highlights corroborated information in green and contradictory or unverified data in orange. This report examines the feature's mechanics, its role in addressing accuracy concerns following the 2025 Super Bowl ad correction, and how its post-response verification style differs from the inline citations found in competing AI assistants.

Microsoft has updated Microsoft 365 Copilot with OpenAI’s GPT-5.2 and introduced support for Anthropic models. These changes, alongside an $80 billion infrastructure investment and in-house chip development, reflect a strategic shift toward a diversified, multi-model platform rather than reliance on a single provider.

The IETF has released RFC 9700 to update OAuth 2.0 security guidance, responding to the growing capabilities of AI agents. This overview explains how the 2025 standards and the in-progress OAuth 2.1 framework aim to secure user data against emerging threats like prompt injection in connected applications.

Google differentiates its artificial intelligence offerings by positioning AI Mode as a search-integrated tool for real-time queries and Gemini as a standalone assistant for creative and multimodal tasks. This report explores their diverging roadmaps, subscription tiers, and functional boundaries within the broader competitive landscape.

A new not-for-profit, Women in AI Australia, has launched to improve female representation in the AI sector. Established by three industry experts, the group aims to provide mentorship and skills training in alignment with the federal government's recently released National AI Plan.

To address a projected shortage of one million professionals in India's growing AI sector, SAS has expanded its Academy for Data & AI Excellence. The updated curriculum features four distinct tracks covering generative AI, data engineering, and business analytics for diverse career stages.

Amid a global surge in demand for artificial intelligence literacy, IBM’s “AI Foundations for Everyone” specialization on Coursera offers a no-code pathway for non-technical professionals. This report examines the four-course curriculum, the context of Coursera's 10 million GenAI enrollments, and how the program compares to competitor offerings from Google and DeepLearning.AI.

OpenAI enters the browser market with Atlas, a Chromium-based tool for macOS featuring a persistent ChatGPT sidebar and autonomous Agent Mode. The release highlights a shift toward AI-mediated browsing but draws scrutiny from security researchers regarding prompt injection risks and data privacy.

Following the Wang Fuk Court fire on November 26, which affected seven towers and resulted in at least 128 casualties , TheDayAfterAI utilized the AI assistant Claude to examine the event through fire science principles. The resulting analysis addresses the public debate regarding the roles of bamboo scaffolding versus plastic safety nets , suggesting that a combination of factors, including the "chimney effect" and specific material properties, contributed to the disaster.

Anthropic releases Claude Opus 4.5 on Nov 24, 2025, featuring "Infinite" context compaction and a 200K token window. The flagship model scores 80.9% on SWE-bench Verified and drops API input pricing to $5 per million tokens, challenging GPT-5.1. Adopted under the ASL-3 safety standard, Anthropic holds a 32% enterprise market share.

Google has updated its open-source Agent Development Kit (ADK) to include fully documented bidirectional streaming capabilities, enabling developers to build real-time, multimodal agents that utilize Gemini Live. Now supporting Python, Java, and Go, the framework allows for low-latency interactions such as user interruptions during voice sessions. This development aligns with broader industry trends, as Gartner forecasts a significant rise in enterprise agent adoption over the coming years.

Google is shifting from a list of blue links to a conversational assistant with the global rollout of AI Mode. Building on AI Overviews — now reaching 1.5 billion users — the feature uses Gemini models to answer complex questions directly. However, the shift faces mounting legal and industry scrutiny over its impact on publisher traffic and the "zero-click" economy.

Launched in late 2025, Grokipedia utilizes the Grok language model to generate over 1 million articles as a proposed alternative to Wikipedia. This report examines the platform’s automated structure and rapid growth alongside independent analyses highlighting its reliance on low-credibility sources and distinct ideological framing compared to traditional human-curated encyclopedias.

xAI has formally introduced Grok 4.1, a new large language model family that includes a dedicated "Thinking" mode designed for complex reasoning tasks. The release, which recorded an initial Elo score of 1483 on the LMArena leaderboard, features a significantly reduced hallucination rate compared to previous versions and is now available to users on the X platform and Grok website.