[Schmidt Sciences](https://www.schmidtsciences.org)

May 26, 2026 (Full proposal deadline, 11:59pm AoE)

$300,000–$1,000,000 per project; project duration 1–3 years

Funds research to detect and mitigate deceptive behaviors in AI models, aiming for practical, generalizable interpretability tools that enhance AI truthfulness and reliability.

Indirect costs capped at 10%; open to global nonprofit research institutions and universities.

--- ## Description This opportunity from Schmidt Sciences supports innovative research to advance interpretability methods for large language models (LLMs), with a focus on detecting and mitigating deceptive behaviors. The program seeks proposals that move beyond academic benchmarks to address real-world risks, developing tools that can identify when AI models provide misleading or harmful advice and interventions that improve model truthfulness. The RFP emphasizes actionable scientific progress, mechanistic insights, and generalizable solutions that enhance the reliability and utility of AI systems. Projects may involve monitoring model reasoning, developing steering methods, and applying these insights to practical AI deployments. --- ## Due Dates - May 26, 2026: Full application/proposal deadline (by 11:59pm Anywhere on Earth) --- ## Funding Amount - Total funding per project: $300,000 to $1,000,000 USD (inclusive of up to 10% indirect costs) - Project duration: 1 to 3 years (budget and duration are independent; e.g., a 1-year project may request up to $1M) - Funding may support personnel, PhD students, postdocs, equipment, software, cloud compute, and other standard research expenses --- ## Eligibility - Open to: - Individual researchers, research teams, and project teams - Universities and research institutions (including international institutions) - Nonprofit research organizations and agencies - Multi-institution collaborations between universities and/or nonprofits - Partnerships with industry or for-profit organizations (for-profits may not lead projects, but may participate as partners) - Not eligible: For-profit entities as lead applicants - Indirect costs must not exceed 10% - All projects must comply with applicable laws and may not include lobbying or political activity - International applicants may be subject to additional due diligence (e.g., equivalency determination via NGOSource) --- ## Application Process - Submit a proposal and budget through the [Schmidt Sciences application portal](https://schmidtsciences.smapply.io/prog/2026_interpretability_rfp/) - Proposals should address the core research agenda: detection and mitigation of deceptive behaviors in LLMs, with an emphasis on practical, generalizable interpretability methods - Applicants may request funding for compute or access to Schmidt Sciences’ computing resources (including GPUs, CPUs, and storage) - No letters of support required or reviewed - Informational webinars: April 2, 2026 and April 28, 2026 (registration links available on the opportunity page) - Reporting requirements: Interim and final reports, plus meetings with program officers --- ## Additional Information - Flexible funding structure: Awards are typically disbursed as a series of one-year gifts or grants; future years contingent on progress - Support may include access to software engineering (via the Virtual Institute for Scientific Software), API credits, and community engagement opportunities - Out-of-scope topics: Research not directly addressing deceptive behaviors in LLMs, general interpretability work without application to deception, or projects focused on policy advocacy --- ## External Links - [Schmidt Sciences opportunity page](https://schmidtsciences.smapply.io/prog/2026_interpretability_rfp/) - [AI Interpretability program overview](https://www.schmidtsciences.org/ai-interpretability/) - [Detailed FAQ](https://docs.google.com/document/d/1z0dfWs6uw3mLCc1yT5mOzJziPMhnuP1-Aj7HpYyDVmU/edit?usp=sharing) - [Research agenda](https://docs.google.com/document/d/1fEaeRbx1A8DtzeV97LnHWLo5lxLHX-tlu-NYxn6bfuo/edit?tab=t.0#bookmark=id.1t1ft5ek2hxt) - [Application portal](https://schmidtsciences.smapply.io/prog/295938/apply/?next=/prog/2026_interpretability_rfp/) --- ## Contact Information | Purpose | Contact | |--------------------------|-------------------------------------------| | General inquiries | [interpretability@schmidtsciences.org](mailto:interpretability@schmidtsciences.org) | | AI program inquiries | [ai@schmidtsciences.org](mailto:ai@schmidtsciences.org) | ---

2026 Interpretability RFP

Sign in to see full grant listing