OpenAI, Google, and Anthropic announced specialised medical AI capabilities within days of each other this month, a clustering that suggests competitive pressure rather than coincidental timing. Yet none of the releases are cleared as medical devices, approved for clinical use, or available for direct patient diagnosis—despite marketing language emphasising healthcare transformation.

OpenAI introduced ChatGPT Health on January 7, allowing US users to connect medical records through partnerships with b.well, Apple Health, Function, and MyFitnessPal. Google released MedGemma 1.5 on January 13, expanding its open medical AI model to interpret three-dimensional CT and MRI scans alongside whole-slide histopathology images. 

Anthropic followed on January 11 with Claude for Healthcare, offering HIPAA-compliant connectors to CMS coverage databases, ICD-10 coding systems, and the National Provider Identifier Registry.

All three companies are targeting the same workflow pain points—prior authorisation reviews, claims processing, clinical documentation—with similar technical approaches but different go-to-market strategies.

Developer platforms, not diagnostic products

The architectural similarities are notable. Each system uses multimodal large language models fine-tuned on medical literature and clinical datasets. Each emphasises privacy protections and regulatory disclaimers. Each positions itself as supporting rather than replacing clinical judgment.

The differences lie in deployment and access models. OpenAI’s ChatGPT Health operates as a consumer-facing service with a waitlist for ChatGPT Free, Plus, and Pro subscribers outside the EEA, Switzerland, and the UK. Google’s MedGemma 1.5 releases as an open model through its Health AI Developer Foundations program, available for download via Hugging Face or deployment through Google Cloud’s Vertex AI. 

Anthropic’s Claude for Healthcare integrates into existing enterprise workflows through Claude for Enterprise, targeting institutional buyers rather than individual consumers. The regulatory positioning is consistent across all three. 

OpenAI states explicitly that Health “is not intended for diagnosis or treatment.” Google positions MedGemma as “starting points for developers to evaluate and adapt to their medical use cases.” Anthropic emphasises that outputs “are not intended to directly inform clinical diagnosis, patient management decisions, treatment recommendations, or any other direct clinical practice applications.”

Benchmark performance vs clinical validation

Medical AI benchmark results improved substantially across all three releases, though the gap between test performance and clinical deployment remains significant. Google reports that MedGemma 1.5 achieved 92.3% accuracy on MedAgentBench, Stanford’s medical agent task completion benchmark, compared to 69.6% for the previous Sonnet 3.5 baseline. 

The model improved by 14 percentage points on MRI disease classification and 3 percentage points on CT findings in internal testing. Anthropic’s Claude Opus 4.5 scored 61.3% on MedCalc medical calculation accuracy tests with Python code execution enabled, and 92.3% on MedAgentBench. 

The company also claims improvements in “honesty evaluations” related to factual hallucinations, though specific metrics were not disclosed. 

OpenAI has not published benchmark comparisons for ChatGPT Health specifically, noting instead that “over 230 million people globally ask health and wellness-related questions on ChatGPT every week” based on de-identified analysis of existing usage patterns.

These benchmarks measure performance on curated test datasets, not clinical outcomes in practice. Medical errors can have life-threatening consequences, translating benchmark accuracy to clinical utility more complex than in other AI application domains.

Regulatory pathway remains unclear

The regulatory framework for these medical AI tools remains ambiguous. In the US, the FDA’s oversight depends on intended use. Software that “supports or provides recommendations to a health care professional about prevention, diagnosis, or treatment of a disease” may require premarket review as a medical device. None of the announced tools has FDA clearance.

Liability questions are similarly unresolved. When Banner Health’s CTO Mike Reagin states that the health system was “drawn to Anthropic’s focus on AI safety,” this addresses technology selection criteria, not legal liability frameworks. 

If a clinician relies on Claude’s prior authorisation analysis and a patient suffers harm from delayed care, existing case law provides limited guidance on responsibility allocation.

Regulatory approaches vary significantly across markets. While the FDA and Europe’s Medical Device Regulation provide established frameworks for software as a medical device, many APAC regulators have not issued specific guidance on generative AI diagnostic tools. 

This regulatory ambiguity affects adoption timelines in markets where healthcare infrastructure gaps might otherwise accelerate implementation—creating a tension between clinical need and regulatory caution.

Administrative workflows, not clinical decisions

Real deployments remain carefully scoped. Novo Nordisk’s Louise Lind Skov, Director of Content Digitalisation, described using Claude for “document and content automation in pharma development,” focused on regulatory submission documents rather than patient diagnosis. 

Taiwan’s National Health Insurance Administration applied MedGemma to extract data from 30,000 pathology reports for policy analysis, not treatment decisions.

The pattern suggests institutional adoption is concentrating on administrative workflows where errors are less immediately dangerous—billing, documentation, protocol drafting—rather than direct clinical decision support where medical AI capabilities would have the most dramatic impact on patient outcomes.

Medical AI capabilities are advancing faster than the institutions deploying them can navigate regulatory, liability, and workflow integration complexities. The technology exists. The US$20 monthly subscription provides access to sophisticated medical reasoning tools. 

Whether that translates to transformed healthcare delivery depends on questions these coordinated announcements leave unaddressed.

See also: AstraZeneca bets on in-house AI to speed up oncology research

Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is part of TechEx and is co-located with other leading technology events. Click here for more information.

AI News is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.

The post AI medical diagnostics race intensifies as OpenAI, Google, and Anthropic launch competing healthcare tools appeared first on AI News.

arXiv:2601.08988v1 Announce Type: new Abstract: Reliable clinical decision support requires medical AI agents capable of safe, multi-step reasoning over structured electronic health records (EHRs). While large language models (LLMs) show promise in healthcare, existing benchmarks inadequately assess performance on action-based tasks involving threshold evaluation, temporal aggregation, and conditional logic. We introduce ART, an Action-based Reasoning clinical Task benchmark for medical AI agents, which mines real-world EHR data to create challenging tasks targeting known reasoning weaknesses. Through analysis of existing benchmarks, we identify three dominant error categories: retrieval failures, aggregation errors, and conditional logic misjudgments. Our four-stage pipeline -- scenario identification, task generation, quality audit, and evaluation -- produces diverse, clinically validated tasks grounded in real patient data. Evaluating GPT-4o-mini and Claude 3.5 Sonnet on 600 tasks shows near-perfect retrieval after prompt refinement, but substantial gaps in aggregation (28--64%) and threshold reasoning (32--38%). By exposing failure modes in action-oriented EHR reasoning, ART advances toward more reliable clinical agents, an essential step for AI systems that reduce cognitive load and administrative burden, supporting workforce capacity in high-demand care settings

arXiv:2601.09072v1 Announce Type: new Abstract: Developing safe, effective, and practically useful clinical prediction models (CPMs) traditionally requires iterative collaboration between clinical experts, data scientists, and informaticists. This process refines the often small but critical details of the model building process, such as which features/patients to include and how clinical categories should be defined. However, this traditional collaboration process is extremely time- and resource-intensive, resulting in only a small fraction of CPMs reaching clinical practice. This challenge intensifies when teams attempt to incorporate unstructured clinical notes, which can contain an enormous number of concepts. To address this challenge, we introduce HACHI, an iterative human-in-the-loop framework that uses AI agents to accelerate the development of fully interpretable CPMs by enabling the exploration of concepts in clinical notes. HACHI alternates between (i) an AI agent rapidly exploring and evaluating candidate concepts in clinical notes and (ii) clinical and domain experts providing feedback to improve the CPM learning process. HACHI defines concepts as simple yes-no questions that are used in linear models, allowing the clinical AI team to transparently review, refine, and validate the CPM learned in each round. In two real-world prediction tasks (acute kidney injury and traumatic brain injury), HACHI outperforms existing approaches, surfaces new clinically relevant concepts not included in commonly-used CPMs, and improves model generalizability across clinical sites and time periods. Furthermore, HACHI reveals the critical role of the clinical AI team, such as directing the AI agent to explore concepts that it had not previously considered, adjusting the granularity of concepts it considers, changing the objective function to better align with the clinical objectives, and identifying issues of data bias and leakage.

arXiv:2601.09152v1 Announce Type: new Abstract: This paper introduces PRA, an AI-agent design for simulating how individual users form privacy concerns in response to real-world news. Moving beyond population-level sentiment analysis, PRA integrates privacy and cognitive theories to simulate user-specific privacy reasoning grounded in personal comment histories and contextual cues. The agent reconstructs each user's "privacy mind", dynamically activates relevant privacy memory through a contextual filter that emulates bounded rationality, and generates synthetic comments reflecting how that user would likely respond to new privacy scenarios. A complementary LLM-as-a-Judge evaluator, calibrated against an established privacy concern taxonomy, quantifies the faithfulness of generated reasoning. Experiments on real-world Hacker News discussions show that \PRA outperforms baseline agents in privacy concern prediction and captures transferable reasoning patterns across domains including AI, e-commerce, and healthcare.

arXiv:2601.08838v1 Announce Type: cross Abstract: Large-scale Text-to-SQL benchmarks such as BIRD typically assume complete and accurate database annotations as well as readily available external knowledge, which fails to reflect common industrial settings where annotations are missing, incomplete, or erroneous. This mismatch substantially limits the real-world applicability of state-of-the-art (SOTA) Text-to-SQL systems. To bridge this gap, we explore a database-centric approach that leverages intrinsic, fine-grained information residing in relational databases to construct missing evidence and improve Text-to-SQL accuracy under annotation-scarce conditions. Our key hypothesis is that when a query requires multi-step reasoning over extensive table information, existing methods often struggle to reliably identify and utilize the truly relevant knowledge. We therefore propose to "cache" query-relevant knowledge on the database side in advance, so that it can be selectively activated at inference time. Based on this idea, we introduce Companion Agents (CA), a new Text-to-SQL paradigm that incorporates a group of agents accompanying database schemas to proactively mine and consolidate hidden inter-table relations, value-domain distributions, statistical regularities, and latent semantic cues before query generation. Experiments on BIRD under the fully missing evidence setting show that CA recovers +4.49 / +4.37 / +14.13 execution accuracy points on RSL-SQL / CHESS / DAIL-SQL, respectively, with larger gains on the Challenging subset +9.65 / +7.58 / +16.71. These improvements stem from CA's automatic database-side mining and evidence construction, suggesting a practical path toward industrial-grade Text-to-SQL deployment without reliance on human-curated evidence.

arXiv:2601.08841v1 Announce Type: cross Abstract: The increasing volume and complexity of scientific literature demand robust methods for organizing and understanding research documents. In this study, we explore how structured knowledge, specifically, subject-predicate-object triples, can enhance the clustering and classification of scientific papers. We propose a modular pipeline that combines unsupervised clustering and supervised classification over multiple document representations: raw abstracts, extracted triples, and hybrid formats that integrate both. Using a filtered arXiv corpus, we extract relational triples from abstracts and construct four text representations, which we embed using four state-of-the-art transformer models: MiniLM, MPNet, SciBERT, and SPECTER. We evaluate the resulting embeddings with KMeans, GMM, and HDBSCAN for unsupervised clustering, and fine-tune classification models for arXiv subject prediction. Our results show that full abstract text yields the most coherent clusters, but that hybrid representations incorporating triples consistently improve classification performance, reaching up to 92.6% accuracy and 0.925 macro-F1. We also find that lightweight sentence encoders (MiniLM, MPNet) outperform domain-specific models (SciBERT, SPECTER) in clustering, while SciBERT excels in structured-input classification. These findings highlight the complementary benefits of combining unstructured text with structured knowledge, offering new insights into knowledge-infused representations for semantic organization of scientific documents.

arXiv:2601.08869v1 Announce Type: cross Abstract: Modern artificial intelligence governance lacks a formal, enforceable mechanism for determining whether a given AI system is legally permitted to operate in a specific domain and jurisdiction. Existing tools such as model cards, audits, and benchmark evaluations provide descriptive information about model behavior and training data but do not produce binding deployment decisions with legal or financial force. This paper introduces the AI Deployment Authorisation Score (ADAS), a machine-readable regulatory framework that evaluates AI systems across five legally and economically grounded dimensions: risk, alignment, externality, control, and auditability. ADAS produces a cryptographically verifiable deployment certificate that regulators, insurers, and infrastructure operators can consume as a license to operate, using public-key verification and transparency mechanisms adapted from secure software supply chain and certificate transparency systems. The paper presents the formal specification, decision logic, evidence model, and policy architecture of ADAS and demonstrates how it operationalizes the European Union Artificial Intelligence Act, United States critical infrastructure governance, and insurance underwriting requirements by compiling statutory and regulatory obligations into machine-executable deployment gates. We argue that deployment-level authorization, rather than model-level evaluation, constitutes the missing institutional layer required for safe, lawful, and economically scalable artificial intelligence.

arXiv:2601.09069v1 Announce Type: cross Abstract: Knowledge graphs (KGs) have commonly been constructed using predefined symbolic relation schemas, typically implemented as categorical relation labels. This design has notable shortcomings: real-world relations are often contextual, nuanced, and sometimes uncertain, and compressing it into discrete relation labels abstracts away critical semantic detail. Nevertheless, symbolic-relation KGs remain widely used because they have been operationally effective and broadly compatible with pre-LLM downstream models and algorithms, in which KG knowledge could be retrieved or encoded into quantified features and embeddings at scale. The emergence of LLMs has reshaped how knowledge is created and consumed. LLMs support scalable synthesis of domain facts directly in concise natural language, and prompting-based inference favors context-rich free-form text over quantified representations. This position paper argues that these changes call for rethinking the representation of relations themselves rather than merely using LLMs to populate conventional schemas more efficiently. We therefore advocate moving from symbolic to natural-language relation descriptions, and we propose hybrid design principles that preserve a minimal structural backbone while enabling more flexible and context-sensitive relational representations.

arXiv:2601.09117v1 Announce Type: cross Abstract: Generative AI systems increasingly enable the production of highly realistic synthetic media. Civitai, a popular community-driven platform for AI-generated content, operates a monetized feature called Bounties, which allows users to commission the generation of content in exchange for payment. To examine how this mechanism is used and what content it incentivizes, we conduct a longitudinal analysis of all publicly available bounty requests collected over a 14-month period following the platform's launch. We find that the bounty marketplace is dominated by tools that let users steer AI models toward content they were not trained to generate. At the same time, requests for content that is "Not Safe For Work" are widespread and have increased steadily over time, now comprising a majority of all bounties. Participation in bounty creation is uneven, with 20% of requesters accounting for roughly half of requests. Requests for "deepfake" - media depicting identifiable real individuals - exhibit a higher concentration than other types of bounties. A nontrivial subset of these requests involves explicit deepfakes despite platform policies prohibiting such content. These bounties disproportionately target female celebrities, revealing a pronounced gender asymmetry in social harm. Together, these findings show how monetized, community-driven generative AI platforms can produce gendered harms, raising questions about consent, governance, and enforcement.

arXiv:2405.10045v3 Announce Type: replace-cross Abstract: This paper presents Global Benchmark Database (GBD), a comprehensive suite of tools for provisioning and sustainably maintaining benchmark instances and their metadata. The availability of benchmark metadata is essential for many tasks in empirical research, e.g., for the data-driven compilation of benchmarks, the domain-specific analysis of runtime experiments, or the instance-specific selection of solvers. In this paper, we introduce the data model of GBD as well as its interfaces and provide examples of how to interact with them. We also demonstrate the integration of custom data sources and explain how to extend GBD with additional problem domains, instance formats and feature extractors.

arXiv:2505.17217v3 Announce Type: replace-cross Abstract: Large Language Models (LLMs) often exhibit gender bias, resulting in unequal treatment of male and female subjects across different contexts. To address this issue, we propose a novel data generation framework that fosters exploratory thinking in LLMs. Our approach prompts models to generate story pairs featuring male and female protagonists in structurally identical, morally ambiguous scenarios, then elicits and compares their moral judgments. When inconsistencies arise, the model is guided to produce balanced, gender-neutral judgments. These story-judgment pairs are used to fine-tune or optimize the models via Direct Preference Optimization (DPO). Experimental results show that our method significantly reduces gender bias while preserving or even enhancing general model capabilities. We will release the code and generated data. We release the code and generated data at: https://github.com/WeiKangda/LLMs-Exploratory-Bias-Mitigation/tree/main.

arXiv:2506.12382v4 Announce Type: replace-cross Abstract: Ensuring the safety and alignment of Large Language Models is a significant challenge with their growing integration into critical applications and societal functions. While prior research has primarily focused on jailbreak attacks, less attention has been given to non-adversarial failures that subtly emerge during benign interactions. We introduce secondary risks a novel class of failure modes marked by harmful or misleading behaviors during benign prompts. Unlike adversarial attacks, these risks stem from imperfect generalization and often evade standard safety mechanisms. To enable systematic evaluation, we introduce two risk primitives verbose response and speculative advice that capture the core failure patterns. Building on these definitions, we propose SecLens, a black-box, multi-objective search framework that efficiently elicits secondary risk behaviors by optimizing task relevance, risk activation, and linguistic plausibility. To support reproducible evaluation, we release SecRiskBench, a benchmark dataset of 650 prompts covering eight diverse real-world risk categories. Experimental results from extensive evaluations on 16 popular models demonstrate that secondary risks are widespread, transferable across models, and modality independent, emphasizing the urgent need for enhanced safety mechanisms to address benign yet harmful LLM behaviors in real-world deployments.

arXiv:2601.08183v2 Announce Type: replace-cross Abstract: Multimodal Large Language Models (MLLMs) show promise in gastroenterology, yet their performance against comprehensive clinical workflows and human benchmarks remains unverified. To systematically evaluate state-of-the-art MLLMs across a panoramic gastrointestinal endoscopy workflow and determine their clinical utility compared with human endoscopists. We constructed GI-Bench, a benchmark encompassing 20 fine-grained lesion categories. Twelve MLLMs were evaluated across a five-stage clinical workflow: anatomical localization, lesion identification, diagnosis, findings description, and management. Model performance was benchmarked against three junior endoscopists and three residency trainees using Macro-F1, mean Intersection-over-Union (mIoU), and multi-dimensional Likert scale. Gemini-3-Pro achieved state-of-the-art performance. In diagnostic reasoning, top-tier models (Macro-F1 0.641) outperformed trainees (0.492) and rivaled junior endoscopists (0.727; p>0.05). However, a critical "spatial grounding bottleneck" persisted; human lesion localization (mIoU >0.506) significantly outperformed the best model (0.345; p

arXiv:2601.08415v2 Announce Type: replace-cross Abstract: Large Language Models (LLMs) are increasingly integrated into academic research pipelines; however, the Terms of Service governing their use remain under-examined. We present a comparative analysis of the Terms of Service of five major LLM providers (Anthropic, DeepSeek, Google, OpenAI, and xAI) collected in November 2025. Our analysis reveals substantial variation in the stringency and specificity of usage restrictions for general users and researchers. We identify specific complexities for researchers in security research, computational social sciences, and psychological studies. We identify `regulatory gray areas' where Terms of Service create uncertainty for legitimate use. We contribute a publicly available resource comparing terms across platforms (OSF) and discuss implications for general users and researchers navigating this evolving landscape.

Nature, Published online: 15 January 2026; doi:10.1038/d41586-026-00035-8

A study in Australia supports genetic screening in young adults before symptoms show, but the generalizability and cost–benefit ratios need to be examined in other settings.

Nature, Published online: 14 January 2026; doi:10.1038/s41586-025-09922-y

Artificial intelligence boosts individual scientists’ output, citations and career progression, but collectively narrows research diversity and reduces collaboration, concentrating work in data-rich areas and potentially limiting broader scientific exploration.