Anwar Ouassini

Harvard AI Expert: Healthcare, LLMs, Agentic AI

Specialties & Experience of this Expert Witness

Additional Information

AI expert witness with 15+ years building, deploying, and auditing 100+ production AI systems. Rare combination of practitioner, scholar, and researcher - the expert opposing counsel can't dismiss as "just an academic" or "just an engineer." Most AI experts fall into one category: academics who've never shipped production code, or engineers who can't write a defensible report. I've done both - built AI systems at Harvard Medical School, Apple, and federal agencies while publishing award-winning peer-reviewed research on AI ethics and large language models. This dual foundation creates opinions that survive Daubert challenges and cross-examination. Technical expertise spans multiple domains, including healthcare AI and insurance bad faith litigation, including prior authorization algorithms, utilization management systems, AI-driven claim denials, and clinical decision support failures. Deep experience with LLM hallucinations and defective outputs including GPT, Claude, and Llama failures, RAG retrieval errors, guardrail inadequacy, and failure to warn claims. Specialized knowledge in agentic AI and autonomous systems covering multi-agent architectures, autonomous decision-making, tool use failures, and human-in-the-loop control breakdowns. Extensive work in algorithmic bias and discrimination involving hiring algorithms, lending models, insurance underwriting, and disparate impact analysis. Additional expertise in AI fraud and misrepresentation including "AI washing," overstated capabilities, and securities disclosure failures. Institutional experience includes Harvard University, Harvard Medical School, Mass General Brigham, Dana-Farber Cancer Institute, Beth Israel Deaconess Medical Center, Boston Children's Hospital, McLean Hospital, FDA, US Department of Defense, US Department of the Interior, University of Chicago, Presbyterian Healthcare, UMass Medical, Molina Healthcare, and Apple. Regulatory expertise covers HIPAA, FDA/SaMD, CMS, FTC Act §5, EEOC/Title VII, SEC, NIST AI RMF, CCPA/GDPR, and State AG Enforcement. Available for insurance bad faith, healthcare AI malpractice, LLM hallucination liability, algorithmic hiring discrimination, product liability, AI fraud, agentic AI liability, securities litigation, regulatory defense, class actions, IP disputes, breach of contract, and wrongful denial of coverage matters. What sets my analysis apart: I understand AI systems from the inside because I've built them. When examining a prior authorization algorithm that denied coverage, I know where to look - training data, decision thresholds, override logic, audit trails. When analyzing an LLM that hallucinated false information, I trace the failure through the retrieval pipeline, evaluate guardrail implementation, and determine if deployment met the standard of care. This isn't theoretical - it comes from building these systems where failures have real consequences. Attorneys receive litigation-ready analysis. Reports connect technical findings to legal standards. I identify documents to request in discovery, formulate technical interrogatories, and anticipate opposing expert arguments. I translate complex AI concepts into language judges and juries understand. I quantify damages using defensible methodologies grounded in technical evidence. My approach emphasizes objectivity. If the AI system performed appropriately, I will say so - my credibility depends on calling it as I see it regardless of which side retained me. Education from Harvard University and University of Chicago in Artificial Intelligence and Computational Methods.