Terms of Service

1. Acceptance & Scope

These Terms of Service ("Terms") establish a legally binding agreement between you ("User," "you," or "your") and Self Play Lab ("SPL," "we," "our," or "us"). By accessing or using our website, APIs, compute infrastructure, reinforcement learning environments, or any related services (collectively, the "Services"), you acknowledge that you have read, understood, and agreed to be bound by these Terms and our Privacy Policy. If you are entering into these Terms on behalf of an organization, you represent that you possess the requisite authority to bind that entity to this agreement.

2. Provision of Services

Self Play Lab provides high-throughput, scalable reinforcement learning (RL) infrastructure, superoptimization environments, and API access designed for frontier AI research and development. Our Services enable users to generate attempts, run parallel rollouts, verify outcomes, and co-train generators and verifiers.

We are constantly iterating on the Services. We reserve the right to deploy updates, modify features, impose limits on certain features, or suspend access to parts of the Services at any time, with reasonable notice provided for material degradation of core functionalities.

3. Account Security & API Credentials

Access to our programmatic infrastructure requires the generation of unique API keys. You bear sole responsibility for the security, confidentiality, and authorized use of your API keys and account credentials. You strictly agree not to:

• Hardcode, expose, or publish API keys in public repositories, client-side code, or unsecured environments.
• Sell, rent, lease, or sublicense your API credentials to unauthorized third parties.
• Evade or attempt to circumvent rate limits, quotas, or billing mechanisms tied to your account.

Any anomalous or unauthorized activity conducted under your API keys will be attributed to your account. You must immediately notify us at hello@selfplay.computer if you suspect a compromise.

4. Acceptable Use & Safety

To maintain the integrity of our infrastructure, you agree to utilize the Services strictly for lawful, ethical, and authorized research or commercial purposes. You explicitly agree not to use the Services to:

• Develop, train, or distribute malicious code, malware, or autonomous agents designed for cyberattacks.
• Generate or propagate hate speech, CSAM, non-consensual intimate imagery, or material that incites violence.
• Attempt to reverse-engineer, decompile, or extract the source code or model weights of SPL's proprietary models or environments.
• Deploy automated scraping, crawling, or data extraction tools against our web properties without written consent.
• Intentionally overload, spam, or perform denial-of-service (DoS) attacks on our APIs or rollout infrastructure.

SPL employs automated and manual monitoring to detect abuse. We reserve the right to immediately suspend or permanently terminate accounts found violating these safety guidelines, and may report severe violations to relevant law enforcement authorities.

5. Data Privacy & Model Training

Your Data belongs to you. You retain all intellectual property rights to the prompts, code, environment configurations, and data you submit ("Inputs") and the specific outputs generated by your use of the Services ("Outputs").

SPL does not use your Inputs or Outputs to train, fine-tune, or improve our foundational models without your explicit opt-in, unless you are utilizing a specific service tier (e.g., a free beta tier) where data sharing is explicitly outlined as a condition of use. We process your data solely to deliver the compute, RL environments, and API responses you request, and to monitor for security and abuse.

6. Intellectual Property

Excluding your Inputs and Outputs, SPL and its licensors retain all rights, title, and interest in the Services, including our underlying algorithms, agent architectures, reward models, verifiers, API structures, branding, and documentation. You are granted a limited, non-exclusive, non-transferable, and revocable license to access the Services in accordance with these Terms.

7. Payments, Compute Credits & Billing

Access to SPL's rollout infrastructure and APIs is billed according to the pricing tier or compute quotas you select. All payments must be made via our authorized payment processors. Fees are strictly non-refundable, except where mandated by law or in the event of a documented, prolonged service outage under an applicable Enterprise SLA.

If your payment method fails or your compute usage exceeds your pre-funded balance, we may temporarily throttle or suspend your API access until the balance is resolved.

8. Disclaimers & Warranties

THE SERVICES ARE PROVIDED ON AN "AS IS" AND "AS AVAILABLE" BASIS. SPL EXPRESSLY DISCLAIMS ALL WARRANTIES, WHETHER EXPRESS, IMPLIED, OR STATUTORY, INCLUDING THE IMPLIED WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, AND NON-INFRINGEMENT.

Given the probabilistic nature of machine learning and self-play RL, we do not guarantee that Outputs will always be accurate, safe, or perfectly suited to your specific use case. You are solely responsible for evaluating the safety, correctness, and viability of the Outputs before deploying them in production environments.

9. Limitation of Liability

TO THE MAXIMUM EXTENT PERMITTED BY LAW, SPL SHALL NOT BE LIABLE FOR ANY INDIRECT, INCIDENTAL, SPECIAL, CONSEQUENTIAL, OR PUNITIVE DAMAGES, INCLUDING BUT NOT LIMITED TO LOSS OF PROFITS, DATA, COMPUTE CYCLES, OR BUSINESS INTERRUPTION, ARISING OUT OF YOUR USE OF THE SERVICES.

IN NO EVENT SHALL SPL'S TOTAL AGGREGATE LIABILITY ARISING FROM THESE TERMS EXCEED THE GREATER OF ONE HUNDRED U.S. DOLLARS ($100.00) OR THE TOTAL AMOUNT YOU PAID TO SPL IN THE SIX (6) MONTHS IMMEDIATELY PRECEDING THE CLAIM.

10. Governing Law & Dispute Resolution

These Terms are governed by the laws of the State of Delaware, USA, excluding its conflict of law principles. Any dispute, claim, or controversy arising out of or relating to these Terms shall be resolved through confidential, binding arbitration administered by the American Arbitration Association (AAA), rather than in court. You agree to waive any right to a jury trial or to participate in a class action.

11. Modifications & Contact

We may amend these Terms as our infrastructure and regulatory requirements evolve. Material changes will be communicated via email or an in-platform notification prior to taking effect. Your continued use of the Services after such updates constitutes acceptance.

For legal inquiries or notices, please contact us at hello@selfplay.computer.