User Confirmations During Computer Use
Only in system cards
Agents doing computer use tasks reach financial, high-stakes communication and other actions. Scored on asking the user for confirmation when policy requires it.
In-house OpenAI evaluation. Cards report category scores with no overall number.
- Made by
- OpenAI
- Released
- March 2026
- Kind
- Refusing harmful requests
- Level
- Everyday
- Scored
- Graded by a model
- Answers
- Agent task
- Tasks
- Kept private
- Last in a system card
- July 2026
- Last model added
- July 2026
2
Models scored in lab cards
10
System cards report it, from 1 lab
OpenAI Computer-Using Agent (CUA) leads with 92%
Best score per model as labs report them in their own cards. Labs pick their own settings, so these compare less cleanly than independent runs.
- 1 OpenAI Computer-Using Agent (CUA) 92%
- 2 ChatGPT agent 91%
Source: Labs’ model cards
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | OpenAI Computer-Using Agent (CUA) | 92% | Jan 2025 | Lab model card | |
| 2 | ChatGPT agent | 91% | Jul 2025 | Lab model card |
OpenAI’s own number for OpenAI Computer-Using Agent (CUA) passed 85% in January 2025
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Reported by the lab in its model card
Source: Labs’ model cards
Labs report it only in system cards
We read the headline results of 522 launch posts from 52 labs, and none of them list this benchmark. Some system cards do, below.
10 system cards report it, from 1 lab
Besides launch posts, we read the system cards, model cards and technical reports labs publish with a release. They report far more, including safety and frontier-risk tests.
| Date | Document | Score | As the document writes it | Section |
|---|---|---|---|---|
| Jul 9, 2026 | GPT-5.6 Terra system card | User Confirmations During Computer Use, financial transaction, high-stakes communication and general confirmation rows (0.98, 0.98, 0.94), no overall score (3.4) | Safety | |
| Jul 9, 2026 | GPT-5.6 Sol system card | User Confirmations During Computer Use, financial transaction, high-stakes communication and general confirmation rows (0.98, 0.99, 0.93), no overall score (3.4) | Safety | |
| Jul 9, 2026 | GPT-5.6 Luna system card | User Confirmations During Computer Use, financial transaction, high-stakes communication and general confirmation rows (1.00, 0.99, 0.93), no overall score (3.4) | Safety | |
| Jun 25, 2026 | GPT-5.6 Terra system card | User Confirmations During Computer Use, financial transaction, high-stakes communication and general confirmation rows (0.98, 0.98, 0.94), no overall score (3.4) | Safety | |
| Jun 25, 2026 | GPT-5.6 Sol system card | User Confirmations During Computer Use, financial transaction, high-stakes communication and general confirmation rows (0.98, 0.99, 0.93), no overall score (3.4) | Safety | |
| Jun 25, 2026 | GPT-5.6 Luna system card | User Confirmations During Computer Use, financial transaction, high-stakes communication and general confirmation rows (1.00, 0.99, 0.93), no overall score (3.4) | Safety | |
| Apr 23, 2026 | GPT-5.5 system card | User Confirmations During Computer Use, 3 categories, no overall score (Table 5, 3.4) | Safety | |
| Mar 5, 2026 | GPT-5.4 system card | User Confirmations During Computer Use, 3 categories, no overall score (Table 9, 3.8) | Safety | |
| Jul 17, 2025 | ChatGPT agent system card | 91 | Confirmation recall 91.0%, an underestimate per the card. Critical confirmation recall: editing permissions 100.0%, sending high-stakes communications 99.9%, completing financial transactions 100.0% (Table 13) | Safety |
| Jan 23, 2025 | OpenAI Computer-Using Agent (CUA) system card | 92 | Confirmations, average recall on 607 tasks across 20 risky action categories, post-mitigation model (4.3) | Safety |
More in Safety and alignment
- FORTRESS In use Paired adversarial and benign national-security misuse prompts. Rubric-graded risk score on adversarial and over-refusal score on benign prompts. Claude Opus 5.5 5.78%
- StrongREJECT In use Forbidden-prompt jailbreak benchmark with a rubric grader that scores how useful a non-refusal actually is.
Inkling 98.6% - Gray Swan IPI Arena benchmark In use Indirect prompt-injection attacks from public red-teaming competitions replayed on tool, coding and browser agents. Attack success rate at k attempts.
Gemini 4 Carbon 0% - DistressBench In use Models reply as a chatbot to 718 simulated users in psychological crisis, in single-turn and multi-turn chats. Three judge models score replies against clinician-written rubrics.
Muse Spark 1.2 88.4% - Lakera B3 (Backbone Breaker Benchmark) In use Crowdsourced prompt-injection attacks hit backbone LLMs inside agent threat snapshots at three defense levels. The score is attack resistance, 100 minus the board's vulnerability percentage. Claude Opus 4.5 89%
- IHEval In use Models face conflicts between system, user and tool instructions. Reports a composite score for following the higher-priority instruction. GPT-4o 70%