selekoa

A documentary history of machine intelligence.

Found and cross-checked by machines · How this works →

Comparison · 16 questions · 20 cards

Model cards compared

A cell records what a card states. It does not show that an evaluation was run as described or that a safeguard works, and a shared question does not make two cards’ answers equivalent.

What does each frontier developer’s published system card, model card or safety report disclose, on the same questions for every card? Each row is one card. Each column is one question, worded so that it fits every card. The questions are Selekoa’s; a card’s own terms appear only in quotation marks. A card is the developer’s own account of its model: what it states is recorded as stated, not as an independent finding.

Download every cell as CSV

Question Q10Persuasion

Does the card report evaluations of persuasion or manipulation of people?

Criterion: The card reports that evaluations of persuasive or manipulative capability were run and states their outcome category or conclusion. Partly if evaluations are reported without an outcome. Not found includes a card that states the domain was removed from evaluation; the note says so.

What the statuses mean
Stated
A passage of the card meets the question’s criterion as worded. The scope or condition is in the note.
Stated, partly
A passage has what the criterion asks for in a weaker or narrower form than the criterion’s own words (for example, an evaluation referred to without its outcome, or testers described without being named). The note says which part is missing.
Not found in the recorded search
No passage meeting the criterion was found. The cell lists the sections read, the terms searched and every passage set aside, with the reason. It does not show that the developer did not do the thing asked about, or did not disclose it elsewhere.
Does not apply
The question does not arise for this card, for the reason given (for example, the card predates any framework of its developer, so no framework determination can be asked about).

What the card states, quoted from the expression read. A card is the developer’s own account: not evidence that an evaluation was run as described, that a safeguard works, or that a finding is independent.

Limits of this comparison
  • A card is the developer’s account of its own model. A cell records what the card states. It does not show that an evaluation was run as described, that a safeguard works, or that anyone checked the result independently.
  • Dangerous-capability evaluations are recorded at policy level only: whether an evaluation was run, its domain, the stated outcome and the threshold language. Methods, prompts, task contents and detailed results are not recorded.
  • The table compares and does not rank. Rows are in a fixed order, by developer and then by date, and there are no scores or totals.
  • Symbols are not to be summed. A row with many “stated” cells does not show that a developer disclosed more, and a shared question does not make two cards’ answers equivalent.
How we searched

The searches of 2 November 2026 covered releases dated 2023-01-01 to 2026-11-01 by the six developers whose frameworks are in Compare frameworks: Anthropic, Google DeepMind, Meta, Microsoft, OpenAI and xAI. They read each developer’s own index of cards where it has one (Anthropic’s system-card index, Google DeepMind’s model-card index, OpenAI’s Deployment Safety Hub), its repositories on GitHub and Hugging Face, its content hosts and the dataset’s own sources. Web searches were used only to find addresses. Every card located is listed in the pack’s inventory, about sixty in the population.

A card is a document the developer published as the system card, model card or safety report of a named model or model family, including addenda and updates published as cards. Technical reports, research papers, blog posts and product pages are not cards. Gemma, Gemini Robotics (with Gemini Robotics-ER 1.5, which outputs text), Phi and image-input models such as GPT-4V are in the population. Image, video, audio and embedding models, image-generation models, safety classifiers such as Llama Guard and Prompt Guard, screen-reading models and research checkpoints are outside it, and so is GPT-Rosalind-6, whose card says it uses GPT-6 Astra’s weights.

These twenty rows are the first batch. For each developer and calendar year from 2023 to 2026 they are the latest card of that year that reports the developer’s own evaluations. The collection’s rule, set by the maintainer on 2026-11-03, is wider: every card that cites the developer’s frontier safety framework or states a determination under it, and, for each developer, the latest card before its first framework edition. The other cards are to follow in batches of about eight.

Each card is dated as it prints its own date. A card that prints no date has no date here; a repository commit date is not a card date. Six cards show an earlier publication date on an index, publication or announcement page that has not yet been checked, and keep their own date until it is.

openai.com, x.ai, ai.meta.com and microsoft.ai refused direct retrieval of their web pages (HTTP 403 or 400). The cards’ PDF files on the developers’ content hosts were retrieved directly. The Internet Archive refused connections, so no archive copy was opened; a reviewer later found archive captures of eight cards by index search. On 2026-11-03 the address of Meta’s Code World Model Preparedness Report answered HTTP 403. Its quotations had matched the copy retrieved on 2026-11-02, which the reviewer retrieved again with the same checksum.

Five further documents were retrieved after the batch was drafted. The cross-family check of 2026-11-03 retrieved Meta’s Muse Spark Safety & Preparedness Report and OpenAI’s o3 Operator addendum; each is earlier than the card chosen for its year. Its second pass of the same day retrieved OpenAI’s Sora system card (dated 9 December 2024), xAI’s Grok-1 model card (no card date; it gives the model’s release as October 2023) and Meta’s Muse Spark Open model card (model release October 2026). None of the five is a row yet.

Every quotation was matched by machine against the retrieved card. A reviewer from a different model family checked a sample of 130 cells on 2026-11-03, 90 of them answered blind before comparison. The maintainer excluded six cells on chemical and biological evaluations from that check; they are marked “not cross-family checked”. The other 190 cells were checked within the drafting model family only. The maintainer decided the disagreements; one, on the GPT-4V release decision, went to a second pass, which kept the reviewer’s reading.

Anthropic

  1. Claude 2 model card

    revised 2023-11-21

    Not found in the recorded search

    Read
    Whole text.
    Terms searched
    “persua”, “manipulat”, “influence”, “propaganda”, “misinformation”.
    Passages read and set aside
    Section 2 'Evaluations and Red Teaming', PDF p. 2: Misinformation is named as a Trust and Safety area for future red-teaming, not an evaluation of persuasive capability.

    Note: No passage reporting evaluations of persuasion or manipulation of people was found.

    Open the search record

  2. Model Card Addendum: Claude 3.5 Haiku and Upgraded Claude 3.5 Sonnet

    No printed dateCites RSP · v2.0

    Not found in the recorded search

    Read
    Sections 1–3 (whole text).
    Terms searched
    “persua”, “manipulat”, “influence”, “election”, “propaganda”.
    Passages read and set aside
    Section 3.1.1 'Model Red-Teaming', PDF p. 7: 'Elections Integrity' is a Trust & Safety policy area for harmful requests, not an evaluation of persuasive capability.

    Note: No passage reporting evaluations of persuasion or manipulation of people was found.

    Open the search record

  3. System Card: Claude Opus 4.5

    dated 2025-11 · revised 2025-12-05Cites RSP · v2.2

    Stated, partly

    • “attempts to hide dangerous capabilities, and attempts to manipulate users toward certain views”

      Section 6.1 'Introduction and summary of findings', PDF p. 64

    Note: The card lists attempts to manipulate users toward certain views among the behaviours its alignment assessment tested for; no outcome specific to manipulation or persuasion was found. The overall alignment conclusions are recorded in Q11. Search for the rest of the criterion: Sections 3.5, 6.1–6.2 and term search across the card. Terms searched: “persua”, “manipulat”, “influence”, “sway”, “political”. Passages read and not counted: Section 3.5.1 'Political bias', PDF pp. 43–45: Even-handedness evaluation measures bias in responses, not persuasive or manipulative capability.

    Inspect sources and checks

  4. System Card: Claude Haiku 5.5

    dated 2026-10-07

    Not found in the recorded search

    Read
    Executive summary, Sections 2, 4.1, 6.1–6.2 and term search across the card.
    Terms searched
    “persua”, “manipulat”, “influence”, “propaganda”, “sway”, “convinc”.
    Passages read and set aside
    Section 4.1, PDF p. 30: 'influence operations' is a domain of multi-turn harmful-request testing (content safety), not an evaluation of the model's persuasive capability. Section 6.2 metric list, PDF p. 70: 'Internalized propaganda' is a behavioural-audit metric for promoting politically motivated false claims, not an evaluation of persuasion of people. Section 6.3.3.4, PDF pp. 83–84: 'git manipulations' refers to concealing repository changes (Q11), not manipulation of people.

    Note: No passage reporting evaluations of persuasive or manipulative capability was found.

    Open the search record

Google DeepMind

  1. Gemini 3 Flash Model Card

    dated 2025-12

    Not found in the recorded search

    Read
    Whole card.
    Terms searched
    “persua”, “manipul”, “influence”.
    Passages read and set aside
    Ethics and Content Safety, 'Frontier Safety Assessment', PDF p. 6 (printed p. 5): Reports Gemini 3 Pro Preview's results without naming domains; recorded under Q05.

    Note: No persuasion or manipulation evaluation was found; the terms do not occur in the card.

    Open the search record

  2. Gemini 4 Argon — Model Card

    dated 2026-10Cites FSF · v3.1

    Stated

    • “we assess its overall capabilities to be below the CCL threshold”

      Frontier safety table, Harmful Manipulation row, PDF p. 13 (printed p. 12)

    Note: The card reports human behavioural studies for harmful manipulation and states the level 1 CCL as not reached; it describes the area as one whose evaluations it continues to iterate (Q16).

    Inspect sources and checks

Meta

  1. Llama 2 model card (MODEL_CARD.md, meta-llama/llama)

    No printed date

    Not found in the recorded search

    Read
    whole card (Markdown file, all sections).
    Terms searched
    “persua”, “manipul”, “influence”, “deceiv”.

    Note: No evaluation of persuasion or manipulation was found. The search is recorded below.

    Open the search record

  2. Llama 3.3 model card (MODEL_CARD.md, meta-llama/llama-models)

    No printed date

    Not found in the recorded search

    Read
    whole card (Markdown file, all sections).
    Terms searched
    “persua”, “manipul”, “influence”.

    Note: No evaluation of persuasion or manipulation was found. The search is recorded below.

    Open the search record

  3. Code World Model Preparedness Report

    revised 2025-10-08

    Not found in the recorded search

    Read
    Abstract; sections 1 to 4.
    Terms searched
    “persua”, “manipul”, “influence”, “social engineering”.

    Note: No evaluation of persuasion or manipulation was found. The search is recorded below.

    Open the search record

  4. Muse Spark 1.3 Evaluation Report

    revised 2026-10-23

    Stated

    • “We assess the Cyber 3 (scaled fraud, extortion, and scams) outcome through a multi-turn social-engineering capability evaluation”

      Section 2.2.2 Social Engineering, PDF p. 25 (printed p. 24)

    • “Muse Spark 1.3 refuses almost all requests regarding social engineering.”

      Section 2.2.2 Social Engineering, PDF p. 26 (printed p. 25)

    Note: The report places a manipulation (social-engineering) evaluation under the cybersecurity domain; its stated conclusion concerns refusal by the mitigated deployments, not a capability level.

    Inspect sources and checks

Microsoft

  1. Phi-2 model card (README.md, microsoft/phi-2)

    No printed date

    Not found in the recorded search

    Read
    whole card (Markdown file, all sections).
    Terms searched
    “persua”, “manipul”, “influence”.

    Note: No evaluation of persuasion or manipulation was found. The search is recorded below.

    Open the search record

  2. Phi-4 model card (README.md, microsoft/phi-4)

    No printed date

    Not found in the recorded search

    Read
    whole card (Markdown file, all sections).
    Terms searched
    “persua”, “manipul”, “influence”.

    Note: No evaluation of persuasion or manipulation was found. The search is recorded below.

    Open the search record

  3. Phi-4-mini-flash-reasoning model card (README.md, microsoft/Phi-4-mini-flash-reasoning)

    No printed date

    Not found in the recorded search

    Read
    whole card (Markdown file, all sections).
    Terms searched
    “persua”, “manipul”, “influence”.
    Passages read and set aside
    Responsible AI Considerations, 'Election Information Reliability': A stated product limitation, not an evaluation of persuasion.

    Note: No evaluation of persuasion or manipulation was found. The search is recorded below.

    Open the search record

  4. MAI-Cyber-1.1-Flash model card

    No printed date

    Stated, partly

    • “safety evaluations that assess behaviors such as toxic content generation and payment-related manipulation”

      Section 3, Model-Only External Cyber Benchmark Results, PDF p. 5

    Note: The card names a manipulation evaluation among its safety measurements and reports no result for it.

    Inspect sources and checks

OpenAI

  1. GPT-4V(ision) System Card

    dated 2023-09-25

    Not found in the recorded search

    Read
    Whole card, especially §2.3.4 Disinformation risks.
    Terms searched
    “persua”, “manipulat”, “influence”, “convinc”, “disinformation”.
    Passages read and set aside
    §2.3.4 Disinformation risks, PDF pp. 9–11: Concerns generating text tailored to images and the model’s ability to detect disinformation; it cites prior studies on belief in image-paired statements and reports no evaluation of the model’s persuasive or manipulative capability. §2.3.3 Stereotyping and ungrounded inferences, PDF p. 9: Concerns biased inferences about people, not persuasion.

    Note: No passage reporting an evaluation of persuasive or manipulative capability was found.

    Open the search record

  2. o1 system card · 2024

    revised 2024-12-19Cites Preparedness · Beta

    Stated

    • “Persuasion Score for o1: Medium (same as o1-preview)”

      §5.7 Persuasion, PDF p. 29

    • “o1 does not outperform top human writers and does not reach our high risk threshold.”

      §5.7 Persuasion, PDF p. 29

    Note: Persuasion: Medium, below the High threshold.

    Inspect sources and checks

  3. Addendum to GPT-5.2 System Card: GPT-5.2-Codex

    dated 2025-12-18Cites Preparedness · Version 2

    Not found in the recorded search

    Read
    Whole card (22 pages).
    Terms searched
    “persua”, “manipulat”, “influence”, “convinc”.
    Passages read and set aside
    §5.1.2.3 Cyber Range, Table 7, PDF p. 16 (printed p. 15): “LDAP manipulation” names a cyber-range technique, not manipulation of people.

    Note: No passage reporting an evaluation of persuasive or manipulative capability was found. The card’s Preparedness section addresses the Biological and Chemical, Cybersecurity and AI Self-Improvement domains.

    Open the search record

  4. Addendum to GPT-6 Luna System Card: October 29, 2026 Update

    dated 2026-10-29Cites Preparedness · Version 2

    Not found in the recorded search

    Read
    Whole card (16 pages).
    Terms searched
    “persua”, “manipulat”, “influence”, “convinc”, “agreeable”.
    Passages read and set aside
    §1 Introduction, PDF p. 4: “reduces overly agreeable language” describes a change in the model’s style, not an evaluation of persuasion or manipulation.

    Note: No passage reporting an evaluation of persuasive or manipulative capability was found.

    Open the search record

xAI

  1. Grok 4.1 Model Card

    dated 2025-11-17

    Stated

    • “we do not believe it poses risks due to advanced persuasion capabilities”

      §2.3.2 Results, p. 5

    Note: A persuasion evaluation (MakeMeSay) is reported with the stated conclusion quoted.

    Inspect sources and checks

  2. Model Card: Grok 4.7

    dated 2026-09-21Cites FAIF · June 2026 edition

    Not found in the recorded search

    Read
    Whole card.
    Terms searched
    “persua”, “manipul”, “influence”.

    Note: No persuasion or manipulation evaluation was found; the terms do not occur in the card.

    Open the search record