Technology Tales

A security research team recently demonstrated how AI, specifically Claude Mythos, was used to identify a vulnerability in Apple’s M5 chip, highlighting the growing ability of attackers to exploit system weaknesses rapidly. While the exploit required physical access and was subsequently patched, it underscores a broader concern that AI can significantly enhance the capabilities of malicious actors, enabling them to uncover complex security flaws more efficiently. This trend raises alarms for platform providers, as advanced attackers, including those with state-level resources, may leverage AI and emerging technologies like quantum computing to breach even more sophisticated security measures. Quantum computers, expected to break current encryption standards by the late 2020s, pose a significant threat, prompting companies such as Google, Apple and Cisco to invest in post-quantum cryptography and quantum-secure networks. However, legacy systems in critical sectors remain vulnerable, emphasising the need for organisations to upgrade infrastructure and adopt quantum-ready solutions. Users are advised to use secure messaging services, maintain strong passwords and stay informed about evolving threats, as the convergence of AI and quantum technologies may introduce even more complex security challenges in the future.

Running a capable AI model locally on older hardware is more achievable than many people assume, thanks to lightweight open-source tools such as Ollama and OpenCode. Ollama can be installed on Windows, Linux or macOS and allows users to download and run smaller language models directly on their own machines, with the Qwen3.5 4B variant being a practical choice as it requires only around 3.5 GB of RAM. Once Ollama is running, OpenCode can be installed via a single terminal command and connected to the local model to create a simple agentic coding environment.

In practice, this setup is capable of handling straightforward development tasks, such as generating a working Python game complete with score tracking and a terminal interface, without requiring cloud services or expensive hardware. However, the approach does have limitations, as the smaller quantised model can struggle with more complex or multi-step engineering work and may stop mid-task, requiring manual prompting to continue. Overall, the combination represents a genuinely useful low-cost local AI workflow for coding experiments, basic scripting and general research, even if it falls short of more demanding professional use cases.

The cybersecurity profession is experiencing a significant crisis of morale, with nearly half of security specialists globally considering leaving their roles within the next year, driven largely by a stark mismatch between the demands placed on them and the recognition they receive. A Harvey Nash survey of over 3,600 technology professionals found that only 29% of cybersecurity workers received a pay increase in the past year, compared to at least half of professionals in other technology disciplines, with many boards dismissing the need for greater investment on the grounds that no major incident has yet occurred. This invisibility of success, combined with mounting workload pressures from legacy infrastructure and distributed working environments, is fuelling widespread dissatisfaction.

The threat environment is simultaneously intensifying, with AI enabling attackers to exploit vulnerabilities far more rapidly than before, compressing the window between disclosure and mass exploitation from weeks to just days, according to Google Cloud Security research. Criminal and state-sponsored groups are increasingly targeting unpatched third-party software, manipulating developer identities through social engineering and stolen credentials, and even destroying backup capabilities to prevent recovery from ransomware attacks. Mandiant research further highlights that automated hand-offs between attacker groups now occur in under 30 seconds on average, with espionage-related intrusions sometimes going undetected for over 120 days. Despite this bleak landscape, cybersecurity professionals are largely not fearful of AI displacing them, recognising instead that AI increases the need for their expertise rather than diminishing it, and those who combine technical knowledge with strong strategic communication and business awareness are likely to remain highly sought after.

In April 2026, OpenAI introduced workspace agents in ChatGPT, designed to assist teams with complex tasks and workflows by operating within organisational permissions and controls. These agents, powered by Codex, can perform functions such as preparing reports, writing code and managing communications, while running in the cloud to maintain continuity even when users are inactive. They are intended for collaborative use, enabling teams to create, share and refine agents across platforms like ChatGPT and Slack, with examples including lead qualification, product feedback routing and risk management. Available in research preview for various plans, workspace agents offer features such as memory retention, tool integration and approval requirements for sensitive actions, with enterprise users gaining additional governance controls. Early adopters have reported reduced manual effort and improved efficiency, with agents capable of automating tasks like data analysis, document generation and vendor screening. The feature is currently free until May 2026, with credit-based pricing thereafter and further enhancements are planned to expand functionality and integration across business tools.

Collaborative AI systems represent a shift from traditional human-AI interactions, where AI merely executes commands, to environments where both parties actively contribute to decision-making. In scientific research, healthcare and business, AI generates insights, identifies patterns and proposes options, while humans provide context, verify findings and make final decisions.

Examples include AlphaFold’s protein structure predictions, which require human interpretation and JPMorgan’s COiN platform, which streamlines contract analysis but relies on legal teams for review. Similarly, BlackRock’s Aladdin system aids risk analysis but leaves portfolio allocations to human managers.

Effective collaboration depends on tools that reveal their reasoning, such as Elicit for research or GitHub Copilot for coding, allowing users to audit and adjust outputs. Success is measured not only by improved outcomes but also by maintaining human oversight, questioning AI suggestions and periodically working without AI to establish benchmarks. Teams that balance AI’s analytical strengths with human judgement achieve better results, identify errors more quickly and explore options that might otherwise be overlooked, highlighting the need for new skills in evaluating and integrating AI into workflows.

The .claude folder, generated by tools integrating with Claude, serves as a hidden directory storing local state, configuration, cached data and context to maintain consistency across interactions within a project. It contains files such as config.json for settings, memory or context folders for persistent information, agents or tasks for structured workflows, logs for debugging and cache for efficiency. While its presence may seem unobtrusive, removing it resets configurations and context, disrupting workflows and requiring reinitialisation of settings. Proper management includes excluding it from version control, periodically clearing unnecessary files and understanding its role in enabling repeatable, agent-based tasks, ensuring developers avoid unintended disruptions while leveraging its functionality effectively.

Recent enhancements in Google NotebookLM have expanded its capabilities beyond basic summarisation, positioning it as a comprehensive tool for research, synthesis and content creation. Key updates include the ability to make targeted revisions to individual slides within presentation decks using natural language prompts, reducing the need for full regeneration. A new PPTX export feature allows seamless integration of AI-generated slides into standard PowerPoint formats, preserving visual layouts and enabling direct use in corporate environments.

Cinematic video overviews automate the production of animated explainer videos from curated data, simplifying communication with non-technical audiences. Users can now generate reports or slide decks directly within chat threads, maintaining context and nuance from discussions. Additionally, support for EPUB and long-form sources enables analysis of dense technical materials, facilitating cross-referencing and synthesis across extensive documents. These capabilities streamline workflows by minimising manual adjustments, allowing users to transition efficiently from raw data exploration to polished deliverables, whether through presentations, videos, or structured reports.

Google's Antigravity extends beyond its initial coding-focused presentation, offering functionalities that streamline research, knowledge management and task coordination. It includes a browser agent capable of autonomously navigating web content and compiling structured findings, a persistent knowledge base that retains context across sessions and the ability to generate visual UI walkthroughs by capturing application interactions. The platform supports parallel task orchestration through an agent manager, allowing multiple processes to run simultaneously and connects to databases via a natural language interface, enabling users to query complex systems without technical expertise. Designed with an architecture that prioritises autonomous agents and integrated workflows, Antigravity positions itself as a versatile tool that addresses a range of professional needs beyond traditional software development, with features tailored to enhance efficiency in research, documentation and data analysis.

As agentic AI development rapidly matures beyond basic large language model pipelines into multi-agent orchestration, memory management and autonomous task execution, five books stand out in 2026 for practitioners building systems where models take action rather than simply respond. Chip Huyen's AI Engineering offers a production-focused full-stack perspective with particular strength in evaluating non-deterministic, multi-step agent systems. The LLM Engineer's Handbook by Paul Iusztin and Maxime Labonne provides a dense, code-heavy guide to LLMOps pipelines, observability and cost optimisation for production-grade deployments. Jay Alammar and Maarten Grootendorst's Hands-On Large Language Models builds foundational intuition about model behaviour through visual explanations, informing better agent component design. Valentina Alto's Building LLM-Powered Applications takes a hands-on approach to agent memory, tool integration, failure handling and multi-agent collaboration patterns. Finally, Prompt Engineering for Generative AI by James Phoenix and Mike Taylor goes well beyond its title, covering chain-of-thought reasoning, ReAct patterns and systematic prompt debugging frameworks that help diagnose why agents fail in practice. Together, these titles cover complementary layers of the agentic AI stack, making them collectively more valuable than any single volume for teams building reliable production systems.

Pope Leo XIV's encyclical Magnifica Humanitas, issued on 15^th May 2026, addresses the profound moral, social and spiritual challenges posed by artificial intelligence and the broader digital revolution, situating them within the longstanding tradition of Catholic Social Doctrine that began with Leo XIII's Rerum Novarum in 1891. Drawing on biblical imagery, particularly the Tower of Babel and Nehemiah's rebuilding of Jerusalem, the document argues that humanity faces a fundamental choice between a technology-driven culture of power and domination on one hand, and a collaborative, dignity-centred civilisation of love on the other. It affirms the inalienable dignity of every human person as the cornerstone of social life and insists that key principles including the common good, the universal destination of goods, subsidiarity, solidarity and social justice must guide the governance and development of AI systems.

The encyclical raises serious concerns about the concentration of technological and economic power in the hands of a small number of private actors, the erosion of truth in public discourse, the displacement of workers, the exploitation of vulnerable populations through digital supply chains, the use of AI in warfare and the normalisation of conflict. It calls for transparent regulation, robust education, the protection of families and young people, and a renewed commitment to diplomacy and multilateralism. Rejecting both transhumanist visions of human enhancement and a resigned acceptance of inequality or war, the document concludes with an appeal to prayer, shared responsibility and a Eucharistic spirituality rooted in the mystery of the Incarnation.