These are selectively chosen examples of emerging specialisms or newly proposed research frameworks, not five verified discoveries of entirely unprecedented disciplines.
Named research paradigm formulated in 2026 · 21 July 2026 (online)
Agentic genomics
Original authors: Manuel Corpas, Heinner Guio and Segun Fatumo · Cell Genomics
What changed: A clearly articulated 'agentic genomics' paradigm: autonomous agents coordinate multi-step genomic analyses and shift validation into a central research challenge.
Prior work, unresolved questions and novelty assessment
Earlier foundations: Genomic workflows, bioinformatics, automation and AI-assisted analysis existed previously.
Still unresolved: Reproducibility, equitable validation, deployment boundaries and clinical-grade assurance.
Classification: Credible newly articulated subfield; not a wholly unprecedented subject.
Read the original research ↗
New governance framework published in 2026 · 12 August 2026
Agentic profiles for governance
Original authors: Atoosa Kasirzadeh and Iason Gabriel · Nature
What changed: Four-dimensional agentic profiles spanning autonomy, efficacy, goal complexity and generality.
Prior work, unresolved questions and novelty assessment
Earlier foundations: Agent theory, multi-agent systems, AI governance and autonomy research long predate 2026.
Still unresolved: Validation of classifications, applicability across sectors and translation into risk controls.
Classification: New explicit framework within an established cross-disciplinary research area.
Read the original research ↗
2026 task-centred research synthesis · July 2026
AI agents for the science of science
Original authors: Yixuan Liu and Yicheng Zhang · Findings of ACL 2026
What changed: A structured taxonomy of agents as simulations of scientific dynamics versus tools supporting scientific analysis.
Prior work, unresolved questions and novelty assessment
Earlier foundations: Science of science, scientometrics, bibliometrics and multi-agent simulation have a long history.
Still unresolved: Bias, data quality, reliability and reproducible evidence of practical value.
Classification: New synthesis and agent-oriented specialism; the broader field is older.
Read the original research ↗
2026 comparative trust architecture · September 2026
Agent identity landscape
Original authors: Fabrice Rochette and Ariel Gentile · Verana research paper
What changed: Compares portable agent identification, accountable operator, verifiable authorisation and privacy of verification across six approaches.
Prior work, unresolved questions and novelty assessment
Earlier foundations: Digital identity, IAM, PKI, attestations, service identity and multi-agent security existed long before 2026.
Still unresolved: Cross-system adoption, independently verified operator identity, revocation and privacy trade-offs.
Classification: A current comparison and research proposal, not proof that agent identity began in 2026.
Read the original research ↗
2026 literature review and synthesis · 17 October 2026
Autonomous research lifecycle evaluation
Original authors: Yifan Gao, Leyi Yu and Haoyue Li · Journal of Superintelligence
What changed: Survey organising scientific agents across literature, hypotheses, experiments, reporting, review, benchmarks and risks.
Prior work, unresolved questions and novelty assessment
Earlier foundations: Automated science, experiment planning, research software and benchmark evaluation have earlier foundations.
Still unresolved: Whole-process independent scientific replication, risk assurance and alignment between benchmarks and scientific outcomes.
Classification: New survey formulation; not a newly invented entire discipline.
Read the original research ↗