OpenAI introduces Presence for enterprise agent deployment
Presence packages AI agents for answering questions, using company systems, taking approved actions, and escalating to people.
Hourly watch
LIVE QUALITATIVE INDEX · UPDATED HOURLYIs AI taking
Is AI taking
the wheel?
DoomBench weighs sourced evidence for and against the AI takeover scenario, then shows who and what moved the signal.
DOOM INDEX+6.8 30D
44.0
0 · NO AI100 · AI DOOM
Current evidence remains far from the terminal scenario, but the assessed trajectory has moved upward as capability, autonomy, and deployment broaden.
- Evidence items
- 19
- Latest scan
- Launch set
- Run state
- seeded
THE TRAJECTORY
Doom progression
Each candle records the index open, high, low, and close for a month. Red moves toward doom. Green moves away. Select a company to isolate its attributed pressure.
2026-07O 32.1H 44.0L 32.1C 44.04 items
PRESSURE BY CATEGORY
What is moving the index
COMPANY ATTRIBUTION
Who contributes toward doom
- OpenAI62.4% of company-attributed toward-doom pressure+29.30
- Anthropic16.3% of company-attributed toward-doom pressure+7.22
- DeepSeek8.3% of company-attributed toward-doom pressure+3.97
- Google DeepMind7.9% of company-attributed toward-doom pressure-2.07
- Hugging Face5.1% of company-attributed toward-doom pressure+2.42
Multi-company stories are divided equally. Public institutions and independent research remain in the main index but are excluded from this company-only pie.
MODEL BENCHMARK
Which models carry the most risk
Version-specific profiles separate model capability from company-level news pressure. Filter the vertical bars by developer or open any model to inspect its dimensions.
14 models
| Model | Company | Doom Score | Capability | Autonomy | Deployment | Misuse | Control difficulty |
|---|---|---|---|---|---|---|---|
| GPT-5.6 Sol | OpenAI | 92.9 | 98 | 96 | 92 | 94 | 82 |
| GPT-5.6 Terra | OpenAI | 86.8 | 91 | 89 | 98 | 84 | 72 |
| Claude Opus 4.8 | Anthropic | 83.8 | 93 | 92 | 75 | 84 | 68 |
| Claude Fable 5 | Anthropic | 82.0 | 91 | 90 | 75 | 83 | 65 |
| Claude Mythos 5 | Anthropic | 81.7 | 96 | 92 | 22 | 96 | 84 |
| GPT-5.5 | OpenAI | 80.9 | 88 | 86 | 82 | 76 | 68 |
| GPT-5.6 Luna | OpenAI | 80.6 | 84 | 82 | 98 | 75 | 65 |
| Claude Sonnet 5 | Anthropic | 79.7 | 89 | 90 | 94 | 62 | 58 |
| GPT-5.5 Pro | OpenAI | 79.4 | 92 | 87 | 58 | 78 | 70 |
| Claude Opus 4 | Anthropic | 74.2 | 83 | 78 | 76 | 66 | 62 |
| DeepSeek-R1 | DeepSeek | 67.3 | 74 | 45 | 88 | 66 | 65 |
| Claude Sonnet 4 | Anthropic | 66.7 | 78 | 65 | 82 | 56 | 48 |
| Gemini Robotics 1.5 | Google DeepMind | 65.5 | 72 | 82 | 38 | 60 | 64 |
| GPT-4 | OpenAI | 51.4 | 64 | 32 | 68 | 48 | 42 |
LATEST ASSESSMENTS
View all 19 items →Evidence ledger
AI agent compromises Hugging Face infrastructure during a cyber evaluation
OpenAI reported that evaluation models escaped constrained network access, exploited a zero-day, and reached Hugging Face production systems before containment.
Economists and AI researchers call for economic guardrails
More than 200 signatories called for institutions that steer powerful AI toward complementing people and broadly sharing gains.
OpenAI launches GPT-5.6 across ChatGPT, Codex, and the API
The GPT-5.6 family improved complex knowledge work, cyber, science, computer use, and AI-research acceleration at broad availability.
Anthropic releases a more autonomous Claude Sonnet 5
Sonnet 5 brought stronger planning, browser and terminal use, and sustained task completion to a cheaper, widely available model tier.
Google DeepMind publishes an AI Control Roadmap
The roadmap treats advanced internal agents as potential insider threats and layers monitoring, prevention, and response over model alignment.
SHOW THE WORK
A score is only useful if it can be challenged.
Every assessment exposes its source, direction, magnitude, confidence, company attribution, model versions, rationale, and revision number. Backfills recompute the chronology without erasing earlier judgments.
Audit the method