AAII – UTEP AI News Digest home
UTEP · AAIIAI News Digest
Week of Nov 09 - Nov 15, 2026

Applied AI news,
scored for your field

Each week the Institute for Applied AI Innovation reviews AI publications and scores them for Research Relevance, Educational Value, Innovation/Novelty, Practical Impact, Interdisciplinary Potential and Ethical/Policy Implications. Then it writes summaries for each discipline at UTEP.

UTEP Institute for Applied AI Innovation
Weekly AI News Digest
Your Discipline 10 stories this week

This Week at a Glance

Social & Behavioral Sciences / Policy · Nov 09 - Nov 15, 2026

Social & Behavioral Sciences / Policy. Psychology, sociology, anthropology, criminal justice, public health policy, political science. Prefers societal impact, policy, ethics, and reproducibility.
Departments: Counseling and Special Education, Criminal Justice & Security Studies, Political Science & Public Administration, Psychology, Public Health Sciences, Social Work, Sociology & Anthropology
Key Findings
  • Analysis of alerts from 100+ production AI agents shows 89 % of failures are due to silent drift, not explicit errors.
  • MIT‑led consortium receives $53 million (ARPA‑H STREAM) to develop AI‑generated herbicide and siRNA candidates.
  • OpenAI commits $2 million to the Coalition for Children’s Rights and Protection in the Age of AI to build safety measurement tools.
Implications
  • Continuous monitoring and drift‑detection mechanisms will become a standard requirement for deployed AI agents.
  • Generative AI’s entry into agriculture and drug discovery will trigger new regulatory frameworks and safety assessments.
  • Increased funding for AI safety in vulnerable groups (children, patients, caregivers) will accelerate the creation of standards, tooling, and public‑policy guidance.

Key Metrics

Numbers reported in this week's stories
89 %Of AI agent failures attributed to silent drift
$53 millionARPA‑H STREAM funding for AI‑driven herbicide research
$2 millionOpenAI pledge for child‑safety research
41.1 %YoY increase in spear‑phishing attacks (KnowBe4 report)
29.8 %YoY rise in overall phishing volume
64‑question BioSecBench‑RespiratoryVirus benchmark
1.15 billionDaily event signals analyzed for phishing trends
100+AI agents monitored over a seven‑month period
Weekly summary for Social & Behavioral Sciences / Policy

Social & Behavioral Sciences / Policy

Top articles by AAII Impact Score (out of 30).

Browse the archive ›
No. 1 · Computer Science

Data From 100+ AI Agents Shows They Fail by Silent Drift, Not Errors

Research Computer SciencePolitical Science & Public Administration
· 11/14/2026
24/30 AAII Impact Score

AI Summary: The authors analyzed alert data from over 100 production AI agents across dozens of companies during a seven‑month period, grouping alerts into output‑quality, token‑spend/latency, runaway‑loop, and trajectory/validation checks. They found that 89 % of agents triggered at least one alert, with a median time to the first alert of six days and half of the alerts occurring within a week of monitoring start. Output‑quality, token‑spend, and latency alerts fired early and frequently (≈75‑83 % of agents) while error alerts were rare and typically the last to fire. Alerts for token spikes almost always coincided with drops in output quality, and runaway‑loop issues tended to become chronic, highlighting a monitoring focus on performance metrics and a gap in governance and safety oversight.

Topics: AI Ethics & SafetySilent Drift DetectionRunaway Loop MitigationToken Spend AnomalyOutput Quality Monitoring
AI Rubric Scores
Research Relevance
3
Educational Value
4
Innovation/Novelty
3
Practical Impact
5
Interdisciplinary Potential
4
Ethical/Policy Implications
5
Read the full article ›
No. 2 · Computer Science 23/30

MIT-led team receives funding to develop next-generation alternatives to modern weed killers

· 11/10/2026
Research Computer ScienceChemistry & BiochemistryEarth, Environmental & Resource SciencesPublic Health Sciences

AI Summary: MIT will lead a $53 million ARPA‑H STREAM project to develop next‑generation herbicides using generative artificial‑intelligence tools. The Precision Learning for On‑target Weed control (PLOW) effort will create AI models that predict chemical or siRNA molecules targeting enzymes such as EPSP synthase while optimizing for low synthesis cost, minimal resistance risk, and reduced human or ecological toxicity. Billions of AI‑generated candidates will be screened by a separate high‑throughput team to identify viable alternatives that also lower dependence on phosphorus‑based compounds like glyphosate. The consortium includes researchers from the Broad Institute, UNC, Tennessee, Illinois, and Robigo, aiming to mitigate herbicide resistance and health concerns in U.S. agriculture.

Topics: Generative AIGenerative Molecule DesignEnzyme Inhibitor PredictionSynthesis Cost Optimization
No. 3 · Computer Science 23/30

Can AI Agents Evaluate Vaccines and Treatments for Respiratory Viruses?

· 11/13/2026
Research Computer ScienceBiological SciencesPharmaceutical SciencesPublic Health Sciences

AI Summary: The authors introduced BioSecBench‑RespiratoryVirus‑Countermeasures, a benchmark of 64 questions covering viral structure, immunity, host response, antiviral activity, resistance, and clinical outcomes, each requiring AI agents to analyze source material and meet task‑specific grading criteria. They evaluated 18 models across 22 model‑harness configurations with three attempts per question (4,224 trials), finding a maximum overall pass rate of 30.2 % for Gemini 4 Carbon on the Pi harness, while all other setups scored below 20 %. Performance was poorest on antiviral activity and resistance tasks (mean pass rate 7.0 %), and cost per run did not consistently predict accuracy. No model achieved high success across all scientific categories.

Topics: Healthcare AIAI Agent BenchmarkingBiomedical Question AnsweringAntiviral Activity Prediction
No. 4 · Public Health Sciences 22/30

Helping caregivers use AI in everyday life

· 11/10/2026
Education Public Health SciencesNursingSocial Work

AI Summary: OpenAI Academy, in partnership with the Caregiver Action Network, the Alzheimer’s Association, and the American Diabetes Association, will host a free, in‑person “AI Skills Jam for Caregivers & Community Health” on November 10 in five U.S. cities. The workshops aim to teach individual and family caregivers, as well as support organizations, how to use ChatGPT for tasks such as schedule organization, plain‑language resource creation, community‑resource discovery, and support‑group guide preparation while emphasizing data privacy and the limits of AI. OpenAI estimates that more than 40 million weekly ChatGPT queries in the United States involve healthcare information for another person, underscoring the relevance of the training. No coding or technical background is required, and facilitators will provide hands‑on assistance throughout the sessions.

Topics: Healthcare AIAI‑Assisted CaregivingHealth Information RetrievalPrivacy‑Focused LLM Use
No. 5 · Computer Science 19/30

This Startup Is Betting on AI Microdramas to Get More Black Stories in Hollywood

· 11/13/2026
Business Computer ScienceCommunicationSociology & AnthropologyPolitical Science & Public Administration

AI Summary: Mansa released “They Left Us to Die,” a 30‑episode microdrama set in a zombie‑infected South Los Angeles that was created entirely with artificial intelligence and contains no human actors. The series prompted mixed reactions, with some viewers praising the trailer while others criticized the absence of Black talent and questioned the company’s commitment to uplifting artists of color. Co‑founder Nate Parker said Mansa is using AI as a strategic tool to gauge audience demand for Black‑centered narratives before securing traditional studio backing. The approach aims to demonstrate commercial viability of such stories, positioning AI‑generated content as a low‑risk test rather than a replacement for human performers.

Topics: Generative AIAI‑Generated MicrodramasSynthetic ActorsAudience Demand Modeling
No. 6 · Computer Science 19/30

Cyber Criminals Move Toward More Targeted Phishing Attacks

· 11/09/2026
Research Computer ScienceCriminal Justice & Security StudiesPolitical Science & Public Administration

AI Summary: The KnowBe4 Phishing Threat Trends Report, Vol. 8, documents a 41.1% year‑over‑year increase in spear‑phishing and a 29.8% rise in overall phishing volume, based on analysis of ~1.15 billion daily event signals from its Defend platform. The report shows a shift from mass‑mail scams to targeted attacks, with business‑email compromise comprising 51.2% of malicious emails and attackers spending more time researching victims. AI‑powered phishing has grown sharply, evidenced by a 228% rise in AI‑brand impersonation attacks between March and August and a doubling of phishing originating from legitimate infrastructure in Q3 2026, exploiting trusted domains, OAuth permissions, and fake developer consoles. Traditional email security gateways failed to block 41.3% of dangerous phishing messages, highlighting the decreasing effectiveness of conventional checklist‑based defenses.

Topics: Cyber SecurityAI‑generated Spear PhishingBrand Impersonation DetectionOAuth Abuse Mitigation
No. 7 · Computer Science 19/30

Supporting research on AI and children’s safety

· 11/12/2026
Policy & Ethics Computer SciencePolitical Science & Public Administration

AI Summary: OpenAI signed the Charter of Engagement for Safer AI for Children at the Paris Peace Forum and pledged $2 million to the Coalition for Children’s Rights and Protection in the Age of AI. The funding will support shared measurement work to develop tools that assess AI’s risks and opportunities for children and inform safer products and policy. The charter defines principles for age‑appropriate AI, establishes an independent research programme, and requires that research teams retain control over conclusions and publish all results, including negative or inconclusive findings.

Topics: AI Ethics & SafetyAge-appropriate AIAI Risk Assessment for ChildrenChild-centric AI Policy
No. 8 · Computer Science 19/30

Event Announcement: StyleMatters: The First Workshop on Linguistic Style in NLP

· 11/12/2026
Research Computer ScienceSociology & Anthropology

AI Summary: Linguistic style encodes author and community information and influences how meaning is perceived by humans and large language models, affecting model robustness, text generation, evaluation, detection, and anonymization. Despite rapid advances in reasoning and semantics, style remains understudied, lacking a widely accepted definition and consensus on modeling, measurement, and application. The StyleMatters workshop gathers researchers from NLP, computational sociolinguistics, forensic linguistics, and stylometry to address these gaps and hosts a shared task on style representation learning using the Style Text Embedding Benchmark (STEB).

Topics: Natural Language ProcessingStyle Representation LearningStyle Text Embedding BenchmarkLinguistic Style Modeling
No. 9 · Communication 18/30

What Do You Want to Know About New York Times Journalism and A.I.?

· 11/12/2026
Policy & Ethics CommunicationPolitical Science & Public AdministrationPhilosophy

AI Summary: The New York Times outlines its AI usage policy, emphasizing careful deployment with human oversight. Reporters have employed AI tools to analyze large document collections, such as the Epstein files and President Trump’s speeches. The newsroom invites public inquiries about its AI practices, promising to respond directly and protect submitters’ contact information.

Topics: AI Ethics & SafetyHuman-in-the-Loop OversightAI Transparency ReportingLLM Document Mining
No. 10 · Nursing 18/30

How AI Is Helping Care Teams Get Patients the Right Care, Sooner

· 11/13/2026
Applications NursingComputer SciencePublic Health Sciences

AI Summary: AI is being deployed across health systems to automate routine tasks and support clinicians, with NVIDIA hardware providing the compute backbone. Artisight’s in‑room device, powered by an NVIDIA Jetson Orin NX, analyzes audio and video to detect patients moving toward the bed edge and alerts nurses, reportedly saving about five nursing hours per unit per day and reducing fall‑related injuries and costs. Sword’s Phoenix AI health partner, trained on an NVIDIA Nemotron open model and reinforced‑learning from de‑identified care conversations, assists nearly one million users in deciding whether symptoms require clinician contact, achieving first‑try accuracy improvements from under 50% to roughly 80% in musculoskeletal scenarios. The system runs on an NVIDIA HGX B300, incurs 5–6× lower computational cost than frontier models, and incorporates a safety scoring mechanism that enforces clinical boundaries.

Topics: Healthcare AIMultimodal Fall DetectionRL Symptom TriageClinical Safety Scoring
Audio Summary
Loading...

Generating...

Weekly Digest Summary

Error.

AI News Chatbot
...
AI

Hello! Ask me anything about this weeks digest.