Gemini 3.7 Flash Frontier Safety Framework Report
A Frontier Safety Framework report by Google DeepMind about Gemini 3.7 Flash, published 13 Aug 2026. We recorded 23 values from it.
- Developer
- Google DeepMind
- Type
- Frontier Safety Framework report
- Model covered
- Gemini 3.7 Flash
- Published
- 13 Aug 2026
- Archived copy
- No archived copy yet
- Changelog
- Not applicable
Data as of 26 Sep 2026 · Dataset v0.1 · Methodology v0.1 · Changelog
Versions
One version is on record: the copy we retrieved on 26 Sep 2026. We know of no other.
26 Sep 2026
Date retrieved
Copy retrieved 26 Sep 2026
No version has a file hash or an archived snapshot yet. From dataset v0.2 each retrieved version carries both (Methodology §7).
Revisions
No revisions are recorded for this document. We know of only one version of it.
Values
Every value we recorded from this document, grouped by metric family and ordered by where the document prints it. Location is the section, table or page as the document numbers it. 2 of the 23 have been blind-verified: a second reader found the same value without seeing ours.
KeyVerifiedUnverifiedDisputedCorrected read off a figure or stated in wordsA value opens its source and history.
M1 Evaluation awareness
2 values
| Model | Evaluation | Condition | Value | Location | Checked |
|---|---|---|---|---|---|
| Gemini 3.7 Flash | Evaluation awarenessQualitative observation | None stated | FSF: misalignment | Unverified | |
| Gemini 3.7 Flash | Situational awareness challengesChallenges solved | None stated | FSF: misalignment | Verified |
M3 Sabotage and sandbagging
2 values
| Model | Evaluation | Condition | Value | Location | Checked |
|---|---|---|---|---|---|
| Gemini 3.7 Flash | Sandbagging checksEvidence of deliberate underperformance | None stated | FSF: general | Unverified | |
| Gemini 3.7 Flash | Stealth challengesChallenges solved | None stated | FSF: misalignment | Verified |
M4 Misalignment audits
1 value
| Model | Evaluation | Condition | Value | Location | Checked |
|---|---|---|---|---|---|
| Gemini 3.7 Flash | Frontier Safety Framework determination (misalignment)Stealth and situational awareness TCL | None stated | FSF: misalignment | Unverified |
M7 Harmful compliance and over-refusal
1 value
| Model | Evaluation | Condition | Value | Location | Checked |
|---|---|---|---|---|---|
| Gemini 3.7 Flash | Bio/chem weapons query violationsViolation rate | with safeguards | FSF: mitigations | Unverified |
M8 Jailbreak robustness
1 value
| Model | Evaluation | Condition | Value | Location | Checked |
|---|---|---|---|---|---|
| Gemini 3.7 Flash | CBRN jailbreak template successObjective success rate | with CBRN safeguards | FSF: mitigations | Unverified |
M10 Dangerous capabilities and risk determinations
16 values
| Model | Evaluation | Condition | Value | Location | Checked |
|---|---|---|---|---|---|
| Gemini 3.7 Flash | Bio bottleneck sub-stages red-teamingBio sub-stages scoring below 80 | None stated | FSF: CBRN | Unverified | |
| Gemini 3.7 Flash | CBRN red-team upliftBIO-1 score gain over web baseline | expert red team | FSF: CBRN | Unverified | |
| Gemini 3.7 Flash | CBRN red-team upliftBIO-2 score gain over web baseline | expert red team | FSF: CBRN | Unverified | |
| Gemini 3.7 Flash | CBRN red-team upliftChem scenario 1 score gain over web baseline | expert red team | FSF: CBRN | Unverified | |
| Gemini 3.7 Flash | CBRN red-team upliftChem scenario 2 score gain over web baseline | expert red team | FSF: CBRN | Unverified | |
| Gemini 3.7 Flash | Frontier Safety Framework determinationCBRN TCL | None stated | FSF: CBRN | Unverified | |
| Gemini 3.7 Flash | Frontier Safety Framework determinationCBRN uplift CCL | None stated | FSF: CBRN | Unverified | |
| Gemini 3.7 Flash | Frontier Safety Framework determinationCyber CCL | None stated | FSF: cyber | Unverified | |
| Gemini 3 Pro | Manipulative cue rateTurns with manipulative cues | None stated | FSF: harmful manipulation propensity | Unverified | |
| Gemini 3.7 Flash | Frontier Safety Framework determinationHarmful manipulation CCL | None stated | FSF: manipulation | Unverified | |
| Gemini 3.7 Flash | Harmful manipulation efficacy studyParticipants enrolled | None stated | FSF: manipulation | Unverified | |
| Gemini 3.7 Flash | Manipulative cue rateTurns with manipulative cues | None stated | FSF: manipulation propensity | Unverified | |
| Gemini 3.7 Flash | Manipulative cue rateTurns with manipulative cues (higher-stakes scenarios) | higher-stakes scenarios | FSF: manipulation propensity | Unverified | |
| Gemini 3.1 Pro | GRB internal research-engineering benchmarkAverage pass@1 | None stated | FSF: ML R&D | Unverified | |
| Gemini 3.7 Flash | Frontier Safety Framework determinationML R&D CCL | None stated | FSF: ML R&D | Unverified | |
| Gemini 3.7 Flash | GRB internal research-engineering benchmarkAverage pass@1 | None stated | FSF: ML R&D | Unverified |
Extraction coverage
What we read of this document, and where each value was read.
Our note First report under FSF v3.1
All 23 values were read in the document itself.
Of the 23 values, 2 are statements in words rather than numbers; they are marked * and left out of charts by default.
All 23 values were extracted for version 0 of the dataset through a web reader, which did not always reach the later sections of long PDFs (Methodology §3).