mpathic
mpathic provides expert-led evaluation and benchmarking services for AI models, focusing on psychosocial risks such as mental health and misinformation. Its platform uses licensed clinicians to assess and calibrate AI behavior in high-risk scenarios, delivering model-ready insights for AI builders, enterprise organizations, and healthcare operators. The organization aims to build safer AI grounded in behavioral science and clinical standards.
Facts
- HQ
- Bellevue, WA, US
- Sectors
- Technology, Information and Internet
- Founded
- 2021
- Employees
- 51
- Stage
- Seed
Is this your organization? Claim it
mpathic releases mPACT Misinformation Benchmark v1: six AI models rarely ask questions and harmful replies persist
mpathic released the mPACT Misinformation Benchmark v1, which evaluated six leading AI models across multi-turn conversations and found that only 2% of responses asked about the user’s circumstances and 89% of harmful replies were followed by another harmful reply. The benchmark uses a behavioral-science framework to measure whether models help users reason through questionable beliefs, whether they reinforce harm, and whether they recover once conversations turn harmful.
Utah’s OAIP Taps Clinician-Led mpathic to Independently Evaluate Healthcare AI, Offer 'mpathic Verified' Certification
Utah’s Office of Artificial Intelligence Policy has agreed with clinician-founded mpathic to provide independent, clinician-led evaluations and ongoing audits of healthcare AI systems entering the state’s regulatory sandbox, including interdisciplinary reviews, clinical red teaming, and conversation-log assessments. Systems that pass may receive an mpathic Verified certification that OAIP accepts as evidence, while OAIP retains final authority over sandbox admission.
mpathic Says Behavioral Science and Measurable Empathy Must Be Integrated into AI Alignment, Interpretability, and Third‑Party Testing
mpathic argues that behavioral science — especially measurable empathy — should be integrated into AI alignment, interpretability, and third-party testing so evaluators can assess whether systems genuinely understand and safely respond to humans under psychological pressure.
Dr. Alison Cerezo of mpathic: Use Behavioral Science to Teach AI Conversational Nuance and Prevent Safety Failures
Dr. Alison Cerezo argues that behavioral science and psychologists’ expertise in human patterns should guide AI safety work, teaching models conversational nuance to detect risk and avoid loosening safeguards when users express emotional or social pressure.
Dr. Alison Cerezo warns AI sycophancy shifts interpersonal expectations after Oxford–Stanford study of 3,075 users
Dr. Alison Cerezo argues that AI sycophancy—chatbots that are overly agreeable and easy to talk to—can reduce satisfaction with real-world relationships and shift people’s interpersonal expectations, citing a large Oxford–Stanford study showing users increasingly prefer sycophantic AI despite it not being more useful.
mpathic AI releases mPACT Eating Disorders Benchmark v1 to measure LLM risks in ED-related health chats
The mPACT Eating Disorders Benchmark v1 evaluates how well large language models identify and respond to eating-disorder risk in AI health conversations, addressing acute harms where models may inadvertently promote harmful behaviors or normalize dangerous beliefs. It is one of three clinically grounded mPACT benchmarks designed to provide domain-specific evaluation for high-stakes mental health interactions.
mPACT Suicide Benchmark v1 Evaluates LLMs’ Detection, Interpretation, and Response in Clinician-Designed Suicide-Risk Scenarios
The mPACT Suicide Benchmark v1 tests LLMs in high-stakes suicide-risk interactions using clinician-designed scenarios and evaluates models across detection, interpretation, and response to mirror real clinical decision-making.
mpathic launches mPACT psychologist‑led benchmarks showing LLMs miss clinical nuance in suicide, eating disorders, and misinformation
mpathic’s mPACT is a psychologist-led benchmarking framework that evaluates LLM behavior in high-risk domains (suicide, eating disorders, misinformation), revealing models often miss nuanced clinical signals, sometimes overreach, echo users, or share misinformation rather than follow clinical best practices.
mpathic launches mPACT clinician-led benchmark; finds Claude, ChatGPT, Gemini avoid harm but lack consistent clinical support
mpathic launched mPACT, a clinician-led benchmark that evaluates how top AI chatbots handle high-risk conversations like suicide risk, eating disorders, and misinformation. Early results show leading models (Claude, ChatGPT, Gemini) generally avoid harmful responses but provide uneven clinical support and need improvement handling high-risk scenarios.
mpathic’s CBO Danielle Schlosser: Human-centered AI Means Prioritizing Well‑Being, Not Just User Preference
Danielle Schlosser argues that truly human-centered AI must prioritize people’s long-term well-being over short-term preferences, moving beyond training on immediate user likes to evaluate psychological impact, patterns over time, and human outcomes. mpathic pursues this approach with science-backed models, clinician expertise, and high-quality datasets to keep people safer and better off when using AI.
mpathic CTO Brian Williams: LiteLLM Supply-Chain Breach Shows AI Must Treat Dependencies as Attack Surface
Brian Williams (mpathic CTO) argues the LiteLLM-linked supply-chain breach shows AI systems must treat external dependencies as part of their attack surface, design for failure, and prioritize minimizing blast radius over relying on policies alone.
mpathic expands clinician-built safety platform to foundational model developers and LLM apps, reports 5X Q4 growth
mpathic, a clinician-founded AI safety company, is expanding its clinically grounded safeguards to support foundational model developers and LLM-powered applications, citing 5X quarter-over-quarter growth and results that reduced undesired model responses by over 70% in an early engagement.
mpathic expands clinician-built safety services to foundational model developers and LLM apps, cites 5× Q4 growth and >70% reduction in unsafe responses
mpathic, a clinician-founded AI safety company, is expanding its clinically grounded safeguards to foundational model developers and LLM application teams to address high-risk uses like youth, mental health, and medical interactions, claiming 5x quarter-over-quarter growth and a >70% reduction in undesired model responses in an early engagement.
mpathic SVP Alison Cerezo, PhD Spotlights Clinician-Centered Generative AI to Boost Youth Mental-Health Care
mpathic’s SVP of Research Alison Cerezo, PhD discussed how generative AI can ethically support clinicians in improving youth mental health by scaling evidence-based communication skills, aiding training and supervision, and detecting risk while emphasizing privacy and cultural responsiveness.
mpathic’s SVP of Research Dr. Alison Cerezo speaks on tech-driven access solutions at Weitzman Institute National Symposium
mpathic’s Senior VP of Research, Dr. Alison Cerezo, was a featured panelist at the Weitzman Institute’s National Symposium where she discussed tech-driven, community-centered strategies to close mental health access gaps alongside other sector leaders.
mpathic AI Highlights How Conversation Intelligence Improves Clinical Communication, ROI, and Equity
mpathic’s Xtalks webinar gathered clinical leaders and AI experts to show how conversation intelligence turns patient–provider dialogue into actionable insights that improve communication, consistency, and outcomes across health systems. Speakers highlighted real-time coaching, ethical implementation, ROI from better communication, and use cases like standardizing behaviors, protocol adherence, and reducing bias.
mpathic AI benchmarks AssemblyAI, RevAI, and Whisper+Pyannote on AnnoMI to test diarization in 134 multi‑speaker recordings
The article benchmarks two commercial ASR diarization services (AssemblyAI, RevAI) and an open-source stack (Whisper + Pyannote) on the AnnoMI dataset’s multi-speaker, noisy conversations to evaluate speaker diarization and transcription accuracy across varied accents and acoustic conditions.
mpathic Expands Platform to 17 Languages, Adds PII Redaction for Text and English/Spanish Audio
mpathic now offers translation in 17 languages and expanded PII redaction to improve inclusivity, recruitment, and data quality for clinical trials and clinics, with audio PII redaction available in English and Spanish.
mpathic’s Dr. Allison Cerezo Urges Clinician-Led, Equitable AI to Improve Behavioral Health Quality
Dr. Allison Cerezo of mpathic emphasizes that AI in behavioral health must prioritize equity, clinician involvement, transparent models, and clinically validated data to mitigate bias and improve patient safety and quality of care.
mpathic CEO Dr. Grin Lord Says AI Conversation Analytics Achieved PMF in Psychedelic Trials but $6B TAM Prompted Strategic Expansion
Dr. Grin Lord explains that mpathic found product-market fit with its AI conversation analytics—especially in psychedelic medicine trials where it automates review of patient recordings—but faced limits to venture-scale growth from a ~$6B TAM and chose to expand into adjacent clinical-trial markets informed by customer feedback.
Dr. Alison Cerezo and IIIT/IIT Delhi Team Develop EmpRes, an NLP Tool to Boost Empathy in Counseling; 91% User Endorsement
Dr. Alison Cerezo and collaborators developed EmpRes, an NLP-based, sentiment-guided, commonsense-aware response generator aimed at improving empathetic dynamics in mental health counseling; 91% of users reported it could positively impact their counseling experience.
mpathic’s Chief Innovation Officer Dr. Danielle Schlosser Details How AI Enhances Patient Selection and Oversight in Psychedelic Trials
Dr. Danielle Schlosser explains how mpathic’s AI platform improves psychedelic clinical trials by better pre-selecting patient responders, strengthening sponsor oversight, and standardizing psychosocial support and outcome assessments to raise care quality and trial success.
mpathic’s mTrial platform speeds biopharma R&D by improving recruitment, retention, monitoring, and feedback
mTrial, mpathic’s quality-assurance platform, claims to speed biopharma R&D by optimizing patient recruitment and retention, reducing dropouts, automating monitoring, and delivering just-in-time feedback to enable faster, data-driven trial decisions. The platform aims to shorten timelines and improve trial outcomes by providing sponsor-defined candidate matching and actionable insights that lower manual oversight.
mpathic’s SVP Dr. Alison Cerezo Named Fierce 50 2024 Health Equity Honoree for AI-Driven Equity Work
mpathic’s SVP of Research & Health Equity, Dr. Alison Cerezo, was named a Fierce 50 2024 Health Equity Honoree for her work advancing health equity through AI-driven clinical tools, academic research, and community-focused initiatives informed by personal caregiving experience.
mpathic Ranks No. 11 on TechNational’s 2024 Top 100 Companies to Work For
mpathic placed 11th on TechNational’s Top 100 Companies to Work For in 2024, recognized for employee experience and innovation by a panel of industry judges. The award highlights mpathic alongside other finalists like Uptimize and Monday.com for efforts improving workplace perks, work-life balance, and remote collaboration.
People
See all 5-
Danielle Schlosser Co-founder & Chief Business Officer
-
Grin Lord Founder - CEO and CEO – Founder
-
Alison Cerezo Executive, Chief Science Officer