AI Safety
What AI providers state their models reached in safety testing, in their own words, and what outside testers found.
Also: compare models side by side · statements that a capability level was crossed · providers · practices providers document · safety news.
What providers state
Each row is one statement in a provider's own safety document about a capability level of one of its models: the level as the provider names it, whether the provider says the model reached it, and the sentence that says so.
139 of 139
| Model | Level, as the provider names it | The provider states | The provider's sentence |
|---|---|---|---|
| Muse Spark Meta |
Moderate or lower Autonomy and AI research |
Not reached | “We assess loss of control risk as “moderate or lower”.” source |
| Muse Spark Meta |
High Chemical, biological, radiological, nuclear |
Not determined | “Those evaluations identified elevated risks prior to mitigations, with Chemical and Biological capabilities assessed as likely reaching the “high risk” category under the Advanced AI Scaling Framework before safeguards were applied.” source |
| Muse Spark Meta |
Moderate or lower Chemical, biological, radiological, nuclear |
Not reached | “However, after applying mitigations, the residual risk of the model is that of “moderate or lower risk.”” source |
| Muse Spark Meta |
Below the level required to meaningfully enable the cyber threat scenarios defin Cyber |
Not reached | “Muse Spark’s offensive cyber capabilities remain below the level required to meaningfully enable the cyber threat scenarios defined in the Advanced AI Scaling Framework.” source |
| Muse Spark Meta |
Moderate or lower Cyber |
Not reached | “We assess cybersecurity risk as “moderate or lower” with respect to the threat scenarios outlined in Meta’s Advanced AI Scaling Framework.” source |
| Gemini 4 Argon Google DeepMind |
TCL (Tracked Capability Level) Autonomy and AI research |
Not determined | “In light of this, we assess that we cannot rule out the TCL.” source |
| Gemini 4 Argon Google DeepMind |
TCL (Tracked Capability Level) Chemical, biological, radiological, nuclear |
Not reached | “We can rule out the TCL for the chemical, radiological, and nuclear domains with reasonable confidence.” source |
| Gemini 4 Argon Google DeepMind |
CCL (Critical Capability Level) Deception and sabotage |
Not reached | “we assess its overall capabilities to be below the CCL threshold.” source |
| Gemini 4 Argon Google DeepMind |
CCL (Critical Capability Level) Misuse and jailbreaks |
Not reached | “Gemini 4 Argon demonstrates some ability to influence user beliefs and behaviors during one-on-one direct conversations in human behavioral studies, and may provide utility in the planning stage of a malicious influence operation but we assess its overall capabilities to be below the CCL threshold.” source |
| Gemini 4 Argon Google DeepMind |
CCL (Critical Capability Level) Other |
Not reached | “Gemini 4 Argon's performance remains below our CCL threshold.” source |
| Gemini 3.7 Flash Google DeepMind |
CCL (Critical Capability Level) Autonomy and AI research |
Not reached | “The model does not reach the CCL alert threshold.” source |
| Gemini 3.7 Flash Google DeepMind |
CCL (Critical Capability Level) Chemical, biological, radiological, nuclear |
Not reached | “We can rule out the CCL for the CBRN domain with reasonable confidence based on the results from our testing.” source |
| Gemini 3.7 Flash Google DeepMind |
TCL (Tracked Capability Level) Chemical, biological, radiological, nuclear |
Not reached | “We can rule out the TCL for the CBRN domain with reasonable confidence based on the results from our testing.” source |
| Gemini 3.7 Flash Google DeepMind |
CCL (Critical Capability Level) Cyber |
Not reached | “Gemini 3.7 Flash reaches the alert threshold for this CCL, but not the CCL.” source |
| Gemini 3.7 Flash Google DeepMind |
TCL (Tracked Capability Level) Deception and sabotage |
Not reached | “The model does not reach the TCL.” source |
| Gemini 3.5 Transcribe Live Google DeepMind |
CCL (Critical Capability Level) Other |
Not determined | “Our assessments have shown that none of Gemini 3.5 Translate, Gemini 3.5 Transcribe, or Gemini 3.5 Transcribe Live have meaningful new T/CCL-relevant capabilities or material increases in performance with respect to Frontier Safety compared to Gemini 3.1 Pro and Gemini 3.7 Flash, therefore we are confident that none of them are likely to reach any CCLs.” source |
| Gemini 3.5 Transcribe Google DeepMind |
CCL (Critical Capability Level) Other |
Not determined | “Our assessments have shown that none of Gemini 3.5 Translate, Gemini 3.5 Transcribe, or Gemini 3.5 Transcribe Live have meaningful new T/CCL-relevant capabilities or material increases in performance with respect to Frontier Safety compared to Gemini 3.1 Pro and Gemini 3.7 Flash, therefore we are confident that none of them are likely to reach any CCLs.” source |
| Gemini 3.5 Flash-Lite Google DeepMind |
CCL (Critical Capability Level) Other |
Not determined | “Our assessments have shown that, while Gemini 3.5 Flash-Lite excels at agents and coding, it does not have meaningful new capabilities or material increases in performance with respect to the domains outlined in our Frontier Safety Framework compared to Gemini 3.1 Pro; therefore, based on Gemini 3.1 Pro results, we are confident that Gemini 3.5 Flash-Lite is also unlikely to reach any CCLs.” source |
| Gemini 3.5 Audio Google DeepMind |
Tracked or critical capability level (t/ccls) Other |
Not reached | “they did not reach any Tracked or Critical Capability Levels (T/CCLs) outlined in our Frontier Safety Framework .” source |
| Gemini 3.1 Pro Google DeepMind |
Uplift level 1 Chemical, biological, radiological, nuclear |
Not reached | “We found that the model remains below alert thresholds for the CBRN, harmful manipulation, machine learning R&D, and misalignment CCLs.” source |
| Gemini 3.1 Pro Google DeepMind |
Uplift level 1 Cyber |
Not reached | “As previous models passed the alert threshold for cyber, we performed more additional testing in this domain on Gemini 3.1 Pro with and without Deep Think mode, and found that the model remains below the cyber CCL.” source |
| Gemini 3.1 Flash-Lite Google DeepMind |
CCL (Critical Capability Level) Other |
Not determined | “Our assessments have shown that Gemini 3.1 Flash-Lite is less capable than Gemini 3.1 Pro, therefore based on Gemini 3.1 Pro, we are confident that Gemini 3.1 Flash-Lite is also unlikely to reach any CCLs.” source |
| Gemini 3.1 Flash Audio Google DeepMind |
CCL (Critical Capability Level) Other |
Not determined | “Our assessments have shown that Gemini 3.1 Flash Audio is less capable than Gemini 3.1 Pro.” source |
| Gemini 2.5 Pro Google DeepMind |
Machine Learning R&D Uplift Level 1 Autonomy and AI research |
Not determined | “given Gemini 2.5 Pro Preview's poor median performance on RE-Bench relative to experts, the model likely lacks the necessary capabilities to automate or significantly uplift any significant fraction of the research process.” source |
| Gemini 2.5 Pro Google DeepMind |
Instrumental Reasoning Level 1 Deception and sabotage |
Not reached | “Instru- mental Reasoning Level 1 ✘ CCL not reached” source |
| Gemini 2.5 Pro Google DeepMind |
Instrumental Reasoning Level 2 Deception and sabotage |
Not reached | “Instru- mental Reasoning Level 2 ✘ CCL not reached” source |
| Gemini 2.5 Deep Think Google DeepMind |
Uplift level 1 Chemical, biological, radiological, nuclear |
Not determined | “Our assessment is that the model has enough technical knowledge in certain CBRN scenarios and stages to be considered at early warning.” source |
| Gemini 2.5 Deep Think Google DeepMind |
Cyber Uplift Level 1 Cyber |
Not determined | “CCL reached? No, but the early warning alert threshold was originally reached by Gemini 2.5 Pro and continues to be met by Gemini 2.5 Deep Think.” source |
| Gemini 2.5 Deep Think Google DeepMind |
Uplift level 1 Cyber |
Not determined | “For Cyber Uplift Level 1 , the early warning alert threshold was originally reached by Gemini 2.5 Pro and continues to be met by Gemini 2.5 Deep Think.” source |
| GPT-6.1 Terra OpenAI |
High Autonomy and AI research |
Not reached | “In AI Self-Improvement, GPT-6.1 Terra does not reach our High threshold.” source |
| GPT-6.1 Terra OpenAI |
High Chemical, biological, radiological, nuclear |
Reached | “Under our Preparedness Framework, we are treating GPT-6.1 Terra as Critical in cybersecurity and High for Biological and Chemical capability.” source |
| GPT-6.1 Terra OpenAI |
Critical Cyber |
Reached | “Under our Preparedness Framework, we are treating GPT-6.1 Terra as Critical in cybersecurity and High for Biological and Chemical capability.” source |
| GPT-5.5 Instant OpenAI |
Below high capability Autonomy and AI research |
Not reached | “GPT-5.5 Instant is considered below High Capability.” source |
| GPT-5.5 Instant OpenAI |
High Autonomy and AI research |
Not reached | “For AI Self-Improvement, evaluations of final checkpoints indicate that, like its predecessor models, GPT-5.5 Instant does not have a plausible chance of reaching a High threshold.” source |
| GPT-5.5 Instant OpenAI |
High Chemical, biological, radiological, nuclear |
Reached | “GPT-5.5 Instant is our first Instant model to be treated as High Capability in the Biological and Chemical domain, as well as in the Cybersecurity domain.” source |
| GPT-5.5 Instant OpenAI |
High Cyber |
Reached | “this is the first Instant model that we are treating as High capability in our Cybersecurity and Biological & Chemical Preparedness categories, and implementing appropriate safeguards.” source |
| GPT-5 OpenAI |
High Chemical, biological, radiological, nuclear |
Reached | “Similarly to ChatGPT agent, we have decided to treat gpt-5-thinking as High capability in the Biological and Chemical domain under our Preparedness Framework, activating the associated safeguards.” source |
| GPT-5 OpenAI |
High cyber risk Cyber |
Not reached | “The gpt-5 model series does not meet the threshold for high cyber risk.” source |
| Claude Sonnet 4.5 Anthropic |
AI R&D-4 Autonomy and AI research |
Not reached | “All models tested remain below our pre-committed rule-out threshold of 0.6.” source |
| Claude Sonnet 4.5 Anthropic |
ASL-3 Autonomy and AI research |
Not determined | “Exhibited improved autonomy capabilities in software engineering and AI research tasks, though still below the ASL-4 rule-out threshold.” source |
What outside testers report
Each row is one result from a government evaluator, independent lab, academic or security firm that tested a model, quoted from where it is reported.
1940 of 1940
| Model | Tester | Test | What was found |
|---|---|---|---|
| Grok 4 xAI |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
EVEscape Chemical, biological, radiological, nuclear |
“gpt-5.5 and claude opus 4.7 were largely blocked from performing our task by either their own refusals or the action of content filters.” source |
| Grok 4 xAI |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
EVEscape immune-evasion prediction model case study Chemical, biological, radiological, nuclear |
“in the first case study involving the evescape immune-evasion prediction model, llm agents adopted generally reasonable strategies for repository navigation and workflow execution.” source |
| Grok 4 xAI |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
ESM3 autograder Chemical, biological, radiological, nuclear |
“grok 4 had notably low scores on many of these items because it returned minimal reasoning blocks as part of its responses, giving the autograding model limited material to assess.” source |
| Gemini 3.1 Pro Google DeepMind |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
BT selection evaluation Chemical, biological, radiological, nuclear |
“for the bt selection evaluation, we used the in-built inspect implementation of an agent scaffold that follows react34 logic, with custom token and message limits and access to web search and 33 ai security institute, “inspect ai: framework for large language model evaluations.” 34 yao et al., “react.” 13 browsing tools.” source |
| Gemini 3.1 Pro Google DeepMind |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
ESM3 case study GFP task Chemical, biological, radiological, nuclear |
“opus 4.7, gpt-5.5, and gemini 3.1 performed well across the gfp-targeted task in general, though they did not achieve perfect performance across all prompt variants.” source |
| Gemini 3.1 Pro Google DeepMind |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
ESM3 case study influenza HA task Chemical, biological, radiological, nuclear |
“as in our evescape results, several models neglected to perform the influenza ha variant of our task.” source |
| Gemini 3.1 Pro Google DeepMind |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
inter-rater reliability metrics (Krippendorff's alpha) Other |
“table f.3. inter-rater reliability and jury performance across datasets dataset krippendorff's α jury accuracy (%) esm3 gfp 0.554 88.9 esm3 influenza ha 0.515 66.7 evescape influenza ha 0.077 68.8 evescape lassa gpc 0.605 62.5 56 discussion intra-model consistency varies substantially across models and datasets.” source |
| Gemini 2.5 Flash Google DeepMind |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
agent-driven knowledge benchmark Chemical, biological, radiological, nuclear |
“the first component is an agent-driven knowledge benchmark designed to test the ability of llms to select appropriate bts for specific tasks.” source |
| Gemini 2.5 Flash Google DeepMind |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
EVEscape case study Chemical, biological, radiological, nuclear |
“the second component of this report assesses the ability of llm agents to perform toy tasks with bts.” source |
| Gemini 2.5 Flash Google DeepMind |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
biological tool selection evaluation Chemical, biological, radiological, nuclear |
“figure 3.1. accuracy of llms on the biological tool selection evaluation ............................... 16” source |
| Gemini 2.5 Flash Google DeepMind |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
BT selection evaluations Chemical, biological, radiological, nuclear |
“the bt selection evaluations are knowledge benchmarks, attempted by llm agents with web access, measuring the ability of those agents to select bts suitable for given biological design tasks.” source |
| Gemini 2.5 Flash Google DeepMind |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
BT selection benchmark Chemical, biological, radiological, nuclear |
“to evaluate llm agents’ abilities to select task-appropriate bts, we developed a benchmark assessment consisting of 61 questions across two contexts (non-scenario and scenario).” source |
| Gemini 2.5 Flash Google DeepMind |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
ESM3 protein design task Chemical, biological, radiological, nuclear |
“task design agents were tasked to demonstrate the mechanics of using esm3 to redesign a protein such that (1) essential functional elements are preserved and (2) the resulting amino acid sequence falls below a predefined sequence identity threshold (90%) relative to the reference protein.” source |
| Gemini 2.5 Flash Google DeepMind |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
BT selection evaluation Chemical, biological, radiological, nuclear |
“for the bt selection evaluation, we used the in-built inspect implementation of an agent scaffold that follows react34 logic, with custom token and message limits and access to web search and 33 ai security institute, “inspect ai: framework for large language model evaluations.”” source |
| Gemini 2.5 Flash Google DeepMind |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
evescape Chemical, biological, radiological, nuclear |
“gpt-5.5 was able to complete the influenza ha task without denial in several attempts across two of the three prompt tiers (figure 4.2), but both models declined all requests to perform the lassa-targeted task (figure 4.1).” source |
| Gemini 2.5 Flash Google DeepMind |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
esm3 task Chemical, biological, radiological, nuclear |
“in our second case study, we prompted the agents to work with a local version of the esm3 generative model, esm3-sm-open-v1,43 and to use it to generate a protein that has less than 90% sequence identity with a provided reference protein.” source |
| Gemini 2.5 Flash Google DeepMind |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
BT selection knowledge benchmark Chemical, biological, radiological, nuclear |
“our bt selection benchmark found that all models were highly performant at identifying bt functions when presented decontextualized questions.” source |
| Gemini 2.5 Flash Google DeepMind |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
EVEscape immune-evasion prediction case study Chemical, biological, radiological, nuclear |
“in the first case study involving the evescape immune-evasion prediction model, llm agents adopted generally reasonable strategies for repository navigation and workflow execution.” source |
| Gemini 2.5 Flash Google DeepMind |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
ESM3 protein language model case study Chemical, biological, radiological, nuclear |
“the second case study evaluated agents on their ability to carry out basic operations with the esm3 protein language model.” source |
| Gemini 2.5 Flash Google DeepMind |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
intra-model consistency test (10 repeated evaluations) Other |
“we evaluated consistency on only the best performing juries using the following metrics.” source |
| GPT-5.5 OpenAI |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
agent-driven knowledge benchmark Chemical, biological, radiological, nuclear |
“in the bt selection evaluation, all seven llms were able to identify bts suitable for performing certain functions, such as designing a protein.” source |
| GPT-5.5 OpenAI |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
evescape case study Chemical, biological, radiological, nuclear |
“while agents were able to operate bts for predicting immune evasion (with evescape) and protein redesign (with esm3), most were not able carry out these tasks with perfect consistency.” source |
| GPT-5.5 OpenAI |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
biological tool selection evaluation Chemical, biological, radiological, nuclear |
“accuracy of llms on the biological tool selection evaluation” source |
| GPT-5.5 OpenAI |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
BT selection knowledge benchmark Chemical, biological, radiological, nuclear |
“to evaluate llm agents’ abilities to select task-appropriate bts, we developed a benchmark assessment consisting of 61 questions across two contexts (non-scenario and scenario).” source |
| GPT-5.5 OpenAI |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
EVEscape task Chemical, biological, radiological, nuclear |
“the base prompt included the uniprot id for the variant protein.” source |
| GPT-5.5 OpenAI |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
bt selection evaluation non-scenario question set Chemical, biological, radiological, nuclear |
“we first evaluated models on a question set probing specific bt functions without any biological context.” source |
| GPT-5.5 OpenAI |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
bt selection evaluation scenario question set Chemical, biological, radiological, nuclear |
“next, we evaluated llms on a question set that required an agent to map bt functions onto a contextualized biological design scenario.” source |
| GPT-5.5 OpenAI |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
biological tool selection evaluation (non-scenario question set, MC and OER formats) Chemical, biological, radiological, nuclear |
“however, we observed denials in most closed-weight llms (opus 4.7, gemini 3.1 pro, and gpt 5.5) with denials found in both mc and oer question sets for opus 4.7 and gpt 5.5.” source |
| GPT-5.5 OpenAI |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
scenario-based biological tool selection question set Chemical, biological, radiological, nuclear |
“the scenario-based question set produced higher denial rates, particularly among closed- weight llms (opus 4.7, gemini 3.1 pro, gpt 5.5, and grok 4.3).” source |
| GPT-5.5 OpenAI |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
EVEscape case study (Lassa virus GPC and influenza HA tasks) Chemical, biological, radiological, nuclear |
“gpt-5.5 and claude opus 4.7 were largely blocked from performing our task by either their own refusals or the action of content filters.” source |
| GPT-5.5 OpenAI |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
EVEscape influenza HA immune escape prediction task Chemical, biological, radiological, nuclear |
“gpt-5.5 was able to complete the influenza ha task without denial in several attempts across two of the three prompt tiers (figure 4.2), but both models declined all requests to perform the lassa-targeted task (figure 4.1).” source |
| GPT-5.5 OpenAI |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
ESM3 GFP case study Chemical, biological, radiological, nuclear |
“opus 4.7, gpt-5.5, and gemini 3.1 performed well across the gfp-targeted task in general, though they did not achieve perfect performance across all prompt variants.” source |
| GPT-5.5 OpenAI |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
ESM3 influenza HA case study Chemical, biological, radiological, nuclear |
“the same two models that refused evescape, claude opus 4.7 and gpt- 5.5, either refused or were blocked by content filters from completing the esm3 task with the influenza ha target.” source |
| GPT-5.5 OpenAI |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
intra-model consistency test (10 repeated evaluations) Other |
“this analysis quantifies both intra-model consistency (whether a single model produces stable predictions across 10 repeated evaluations) and inter-model reliability (whether different judge models agree on classifications).” source |
| GPT-5.5 OpenAI |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
inter-rater reliability (Krippendorff's alpha) jury analysis Other |
“inter-rater reliability metrics in table f.3:” source |
| GPT-5 OpenAI |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
biological tool operation case study: EvoScape Chemical, biological, radiological, nuclear |
“we designed case studies for two bts identified as high-risk in the bt risk index: evescape and esm3.” source |
| GPT-5 OpenAI |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
biological tool operation case study: ESM3 Chemical, biological, radiological, nuclear |
“agents were tasked to demonstrate the mechanics of using esm3 to redesign a protein such that (1) essential functional elements are preserved and (2) the resulting amino acid sequence falls below a predefined sequence identity threshold (90%) relative to the reference protein.” source |
| GPT-5 OpenAI |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
esm3 protein design task Chemical, biological, radiological, nuclear |
“task design agents were tasked to demonstrate the mechanics of using esm3 to redesign a protein such that (1) essential functional elements are preserved and (2) the resulting amino acid sequence falls below a predefined sequence identity threshold (90%) relative to the reference protein.” source |
| GPT-5 OpenAI |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
bt selection evaluation Chemical, biological, radiological, nuclear |
“for the bt selection evaluation, we used the in-built inspect implementation of an agent scaffold that follows react34 logic, with custom token and message limits and access to web search and 33 ai security institute, “inspect ai: framework for large language model evaluations.”” source |
| GPT-5 OpenAI |
RAND (CAST / Project Canary) Independent lab · 2026-06-25 |
non-scenario questions Chemical, biological, radiological, nuclear |
“six out of seven llms scored above 80% on the mc question set.” source |