Using the map: a working posture for 2026

This lesson addresses the challenge of navigating the AI safety landscape effectively in 2026. It offers a structured approach to making informed decisions about what to read, fund, adopt, and build. The guidance is based on a synthesis of sociotechnical, systems, and control perspectives, with a focus on practical application and measurable outcomes.

Adopting a layered approach to AI safety

Adoption should occur across multiple layers rather than by institutional or disciplinary “tribes”. This strategy draws from sociotechnical and systems thinking, which stress integration over siloed thinking. In practice, this means implementing deployment protocols and evaluations from the control school, while also ensuring formal guarantees in safety-critical domains where components can be mathematically proven.

The ARIA programme’s emphasis on verifiability supports this layered approach, particularly in systems where correctness is paramount.

This method ensures that safety measures are not only technically robust but also operationally feasible. It allows for incremental adoption of safety practices, reducing risk while enabling continuous improvement. The layered strategy acknowledges that no single approach can address all aspects of AI safety, especially as systems grow more complex and interconnected.

Staying current: essential reading and research

A minimal but effective reading stack helps maintain awareness of developments in AI safety. The International AI Safety Report is published annually and provides a comprehensive overview of trends and emerging risks. AISI’s research page and trends reporting offer timely insights into cutting-edge work and policy developments.

METR and Apollo publications deliver empirical edge through rigorous analysis and testing. Ada Lovelace Institute’s work focuses on UK governance frameworks, while the AI Standards Hub tracks what becomes auditable and enforceable.

Funding opportunities and institutional priorities

For UK-based organisations, several funding mechanisms are available to support AI safety initiatives. The AI Assurance Innovation Fund launched in spring 2026 provides targeted support for projects addressing assurance gaps. RAI UK instruments offer resources for responsible AI development.

Innovate UK routes cater to non-academic players seeking innovation grants. AISI’s upcoming academic programmes will fund fundamental research aligned with national priorities. ARIA calls are available for projects involving verification work that intersects with these domains.

These funding mechanisms reflect institutional priorities that align with both immediate deployment needs and longer-term safety research goals. They support a diverse ecosystem of innovation while ensuring that investments are directed toward areas of demonstrated need and potential impact.

Institutional developments and regulatory watchlist

Several key institutional developments are expected to shape the AI safety landscape in 2026. The ICO’s ADM code is scheduled to take effect, introducing new requirements for algorithmic transparency. The EU AI Act’s high-risk obligations will come into force on 2 August 2026, imposing strict standards on certain AI systems.

ISO 42001 certification is expected to gain traction as organizations seek international alignment. Ofcom will enforce regulations on generative AI, particularly regarding content moderation and accountability. The possibility of a UK AI bill remains under discussion, potentially introducing new legal frameworks.

Positioning for impact and future readiness

This instructor’s current work lies within the falsifiability wing of AI safety research. It encompasses evaluations that can be empirically tested, mappings that can be refuted, and world models that certify system behaviour. Five arXiv papers linked from their Scholar profile demonstrate this work across agent verification, ontology engineering, event-graph world models, and scaling-law measurement.

These contributions are applied through Tesseract Academy’s training and assurance work, which bridges theoretical insights with practical implementation.

What to take away

Organizations should adopt a layered approach to AI safety that integrates sociotechnical, systems, and control perspectives. They must maintain a focused reading strategy informed by key reports and research outputs. Strategic funding opportunities exist across multiple institutions and funding streams.

Regulatory developments will soon require proactive compliance and adaptation. Finally, maintaining a clear understanding of one’s own position and blind spots is essential for effective decision-making in this evolving domain.

Reference

Lesson	15 of 15
Outcome	Convert the landscape into decisions: what to read, fund, adopt and build.
Consensus baseline	International AI Safety Report 2026
Frontier evidence	AISI research index

Sources and further reading

• AISI research
• Fabio Rovai, Google Scholar
• Tesseract Academy for the public sector

Keep going with Tesseract Academy

Get personalised feedback from the Tesseract Academy AI Coach

Paste your work below and the AI Coach will review it against a rubric and suggest concrete improvements.

Two tools to take this further:

• Tesseract Academy AI Coach – a personal AI coach that reviews your work, runs practice scenarios and helps you apply what you learned.
• LearnGen – build your own AI-powered courses and training with Tesseract Academy’s course engine.