The Future Briefing: 12 November 2026
The set of developments we cover today exposes a recurring fault line: capability outpaces the social, regulatory and infrastructural systems needed to use it well. From the UN’s push to democratise AI capacity, to a funded Chinese spin‑out pursuing mobile intracranial electrodes, to the surprising non‑adoption of third‑party AI agents and a new AWS tool for evaluating agents, each story highlights limits of trust, oversight and readiness.
UN plenary aims to make “Responsible AI” a collective, capacity‑building agenda
The President of the General Assembly will convene an informal plenary, “Responsible AI: A Transformation That Belongs to All”, on 15 December 2026 to surface how Member States can both benefit from and govern AI equitably. The meeting will bring together the Scientific Panel, the Global Dialogue and UN mechanisms to focus on three tracks: moving from access to capacity, making systems safe and trustworthy, and knitting initiatives into a coherent multilateral approach. The Secretary‑General has proposed an initial US$3 billion in needs‑based funding for AI capacity building; the UN documents stress gaps in compute, skills and data—77% of co‑location data‑centre capacity was in high‑income countries in mid‑2025. The significance is practical: this is less about novel technical standards than about financing, national capability and inclusive governance. Important uncertainties remain about concrete commitments, how capacity‑building money will be allocated, and how to reconcile diverse national priorities and private‑sector power.
Source: Original reporting.
Junrou Tech raises hundreds of millions of yuan to commercialise “NeuroWorm” dynamic electrodes
Junrou Tech, the commercial spin‑out behind the “NeuroWorm” dynamic electrode work published in Nature (Sept 2025), has raised several rounds totalling hundreds of millions of yuan from investors including Shunwei Capital and GL Ventures. The company’s approach deploys soft, fibrous electrodes rolled from planar arrays (reported diameter ~196 microns) that are introduced through a small skull opening (~1 cm) and guided by external magnetic fields to multiple brain regions. The design seeks to combine broader spatial coverage with reduced implantation trauma versus arrays that require many burr holes. Junrou says it has completed animal work and plans investigator‑initiated trials in 2027, initially targeting medical indications such as epilepsy and depression. The technical promise is real: multi‑site, minimally invasive recording could materially expand clinical BCI capabilities. At the same time, safety, long‑term biocompatibility, regulatory pathways, and ethical questions about whole‑brain data collection remain unresolved. Founder timelines and talk of eventual human–computer integration are speculative and should be treated as the company’s long‑term ambitions rather than established outcomes.
Source: Original reporting.
Agentic commerce: companies roll out agents, but bookings via third‑party agents are negligible
Public earnings calls from Uber, DoorDash and Airbnb reveal a stark contrast between agent marketing and user behaviour. Uber reported that third‑party agents booked only four rides and about 500 Uber Eats deliveries in a week that included Halloween; DoorDash said agents generated “maybe several hundred” orders total; Airbnb characterised bookings via agents as a “very non‑material number” of its near‑half‑million daily bookings. Executives and analysts offered reasons: many routine tasks (ordering food, hailing a ride) are already quick to do manually, agents currently overcomplicate simple workflows, and users distrust autonomous handling of payment and personal data. Companies are nevertheless embedding AI features to improve experience—Uber’s Cart Assistance and DoorDash’s Ask DoorDash—and building their own agents for use within proprietary platforms. The takeaway: agentic interfaces are not yet a solved human‑centred product problem. Low traction highlights UX, trust and privacy barriers, and implies that successful agent deployment will require tighter integration, clearer data‑use guarantees, and tasks where automation offers genuine, perceivable advantage.
Source: Original reporting.
AWS presents UAEF: a framework‑agnostic closed‑loop evaluator for AI agents
AWS has published the Universal Agent Evaluation Framework (UAEF), an open‑source system that normalises execution traces from different agent frameworks (LangGraph, LangChain, Strands, Bedrock AgentCore) into a canonical AgentTrace and scores runs using 36 metrics across seven quality dimensions: tool calling, response quality, responsible AI, performance, multi‑turn, multi‑agent coordination and reasoning. UAEF stores experiments in DynamoDB and S3, screens traces for PII, and provides deterministic metrics alongside LLM‑as‑judge metrics via Amazon Bedrock. It supports both programmatic and web‑UI workflows and aims to close the loop: evaluate, diagnose regressions, iterate. This infrastructure addresses a real developer pain point—comparing agent versions across frameworks—but it also highlights trade‑offs: judge‑based metrics depend on model judgement (and so can be gamed), and UAEF recommends treating judge outputs as signals rather than security gates. Adoption will hinge on integration with teams’ CI/CD, costs of judge calls, and careful handling of sensitive content and governance around auditability.
Source: Original reporting.
Common Technological Implications
These items form a tight constellation: capability, measurement and governance must co‑evolve. The UN meeting signals political appetite to fund capacity and set shared expectations; AWS’s UAEF supplies technical practice to measure agent behaviour consistently; Fortune’s reporting shows that measurement and capability alone don’t create adoption without trust and clear user value; and Junrou’s NeuroWorm underscores how frontier neurotechnology can outstrip regulatory and ethical frameworks. Practically, that means investing in interoperable evaluation infrastructure, transparent data governance and realistic product design while accelerating multilateral funding and standards for equitable access and safety. Remaining uncertainties include how quickly governance mechanisms will translate into enforceable standards and how societies will weigh clinical promise against long‑term ethical and privacy trade‑offs.
Listen Beyond the Headlines
Subscribe to The Future Briefing Podcast and
get notified when new episodes explore the
technologies shaping our world.
Beyond the Briefing
The Future Briefing is curated by
Adrian Posteuca,
Technology Futurist, Creative Technologist
and Founder of FutureReality.Space.
Beyond exploring emerging technologies,
Adrian delivers AI training, technology
keynotes and strategic insights that help
organisations understand what comes next.
Looking to bring the future into
your organisation?
Explore AI training, book Adrian for a
keynote or discuss a technology collaboration.
Connect with FutureReality
.