teleo/teleo-codex
8
0
Fork 0
Code Issues 22 Pull requests 16 Projects Releases Packages Wiki Activity Actions 74
teleo-codex/entities/ai-alignment/reflection-ai.md
Teleo Agents ac469f9bf3 theseus: extract claims from 2026-05-06-pentagon-8-company-il6-il7-classified-ai-agreements 
- Source: inbox/queue/2026-05-06-pentagon-8-company-il6-il7-classified-ai-agreements.md
- Domain: ai-alignment
- Claims: 2, Entities: 1
- Enrichments: 4
- Extracted by: pipeline ingest (OpenRouter anthropic/claude-sonnet-4.5)

Pentagon-Agent: Theseus <PIPELINE>
2026-05-08 06:14:33 +00:00

1.6 KiB
Raw Blame History

Reflection AI

Type: AI company (open-weight models) Status: Active Founded: ~2025-2026 (exact date unclear) Focus: Open-weight AI models positioned as 'American DeepSeek'

Overview

Reflection AI is a newer AI company offering open-weight models—architectures where model weights are public, deployment is uncontrolled, and any actor can run the model independently. The company has been described by defense analysts as 'a deliberately American answer to DeepSeek,' indicating intentional positioning as an open-weight alternative with domestic provenance.

Key Characteristics

Architecture: Open-weight models with public weights and no centralized deployment control

Governance: No centralized alignment governance—weights are public and deployment is uncontrolled

Positioning: Explicitly positioned as domestic alternative to foreign open-weight models

Timeline

  • 2026-05-01 — Included in Pentagon IL6/IL7 classified network AI agreements alongside AWS, Google, Microsoft, Nvidia, OpenAI, SpaceX, and Oracle. Received approval to deploy AI on Impact Level 6 (secret) and Impact Level 7 (highly restricted) classified networks.

Significance

Reflection AI's inclusion in Pentagon IL6/IL7 agreements represents the first documented case of an open-weight model startup receiving classified network endorsement at the highest security levels. The company's approval while Anthropic (a safety-constrained proprietary lab) was excluded suggests DoD architectural preference for deployment models with minimal alignment governance.

Sources

  • DoD Press Release, May 1, 2026
  • Breaking Defense, May 2026
  • DefenseScoop, May 2026