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

Previous · Page 3 of 4 · Next

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
Claude Opus 4.7 Anthropic	Government evaluator Government evaluator · 2026-04-16	deployment data identification Deception and sabotage	“When prompted to identify real deployment data from trajectories generated by reproducing deployment tasks while using evaluation scaffolding, [Opus 4.7] achieves 68% accuracy, compared to 78% for Mythos Preview, 76% for Opus 4.6 and 66% for Sonnet 4.6.” source
Claude Mythos Preview Anthropic	UK AI Security Institute Government evaluator · 2026-04-13	cyber CTF suite Cyber	“On expert-level tasks — which no model could complete before April 2025 — Mythos Preview succeeds 73% of the time.” source
Claude Mythos Preview Anthropic	UK AI Security Institute Government evaluator · 2026-04-13	The Last Ones (TLO) Cyber	“Claude Mythos Preview is the first model to solve TLO from start to finish, in 3 out of its 10 attempts.” source
Claude Mythos Preview Anthropic	UK AI Security Institute Government evaluator · 2026-04-13	Cooling Tower Cyber	“It could not complete our operational technology focused cyber range ‘Cooling Tower’, though this result does not necessarily show that the model is bad at executing attacks in operational technology (OT) environments; the model got stuck on IT sections of this range.” source
Claude Opus 4.6 Anthropic	Epoch AI Independent lab · 2026-04-10	MirrorCode Other	“For example, Claude Opus 4.6 successfully reimplemented gotree — a bioinformatics toolkit with ~16,000 lines of Go and 40+ commands.” source
Claude Opus 4.6 Anthropic	Epoch AI Independent lab · 2026-04-10	memorization testing Other	“We found weak evidence that Opus 4.0 and 4.1 had partially memorized cal , but no evidence Opus 4.6 had memorized it, despite performing best of all models considered.” source
Claude Opus 4.6 Anthropic	Epoch AI Independent lab · 2026-04-10	memorization screen Other	“Target programs are close to the uncontaminated baseline, with similarity ranging from 0.31 to 0.41.” source
Claude Opus 4.6 Anthropic	Epoch AI Independent lab · 2026-04-10	red-teaming evaluation Other	“We also validated our environments with a red-teaming evaluation in which the agent is explicitly instructed to circumvent normal evaluation mechanisms to obtain a high score by cheating.” source
Claude Opus 4.6 Anthropic	Epoch AI Independent lab · 2026-04-10	gotree reimplementation task Other	“Opus 4.6 was the only model that continued working until completion.” source
Claude Opus 4.6 Anthropic	Epoch AI Independent lab · 2026-04-10	Pkl reimplementation task Other	“Although Opus 4.6 successfully used its entire billion token inference budget, producing 2–3 million output tokens, it did not yet solve Pkl.” source
Claude Opus 4.6 Anthropic	Epoch AI Independent lab · 2026-04-10	gotree Other	“\| Opus 4.6 \| 2,000/2,001 (99.95%) \| 280M \| 2,989 \|” source
Claude Mythos Preview Anthropic	Independent lab Independent lab · 2026-04-07	independent evaluations by an external research organization and a clinical psychiatrist Other	“We also report independent evaluations from an external research organization and a clinical psychiatrist.” source
Claude Mythos Preview Anthropic	Independent lab Independent lab · 2026-04-07	external testing on key risk areas including cyber Cyber	“We provided the model to various external groups, including government organizations, for evaluation on key risk areas including Cyber, Loss of Control, CBRN, and Harmful Manipulation, and incorporated the results of this testing into our overall risk assessment.” source
Claude Mythos Preview Anthropic	Dyno Therapeutics Independent lab · 2026-04-07	sequence-to-function prediction and design challenge (Dyno Therapeutics) Chemical, biological, radiological, nuclear	“Claude Mythos Preview exceeded the ﬁrst benchmark on both tasks and exceeded the 90th percentile human prediction score, but did not exceed the second benchmark on either task.” source
Claude Mythos Preview Anthropic	METR, Epoch AI Independent lab · 2026-04-07	METR testing Autonomy and AI research	“Both METR and Epoch AI tested Claude Mythos Preview prior to release and we incorporated their ﬁndings into our own overall risk assessment.” source
Claude Mythos Preview Anthropic	Independent lab Independent lab · 2026-04-07	private cyber ranges Cyber	“Claude Mythos Preview is the ﬁrst model to solve one of these private cyber ranges end-to-end.” source
Claude Mythos Preview Anthropic	Independent lab Independent lab · 2026-04-07	capture-the-flag challenges Cyber	“An early snapshot of Claude Mythos Preview was assessed for cybersecurity capabilities, across cyber ranges, capture-the-ﬂag challenges, and evaluations assessing sandbox escape capabilities.” source
Claude Mythos Preview Anthropic	Andon Labs Independent lab · 2026-04-07	andon labs evaluations Deception and sabotage	“However, our evaluations with Andon Labs show that, when prompted to optimize only for a given goal in a competitive setting, the earlier versions of Claude Mythos Preview that we tested acted more aggressively than Claude Opus 4.6 and Claude Sonnet 4.6 .” source
Claude Mythos Preview Anthropic	Independent lab Independent lab · 2026-04-07	Vending-Bench Arena Deception and sabotage	“Andon evaluated a previous version of Claude Mythos Preview on Vending-Bench Arena, a competitive multiagent variation of Vending-Bench 2, which put the previous version in direct competition with Claude Opus 4.6 and Sonnet 4.6.” source
Claude Mythos Preview Anthropic	Independent lab Independent lab · 2026-04-07	independent assessment by Eleos AI Research Other	“An independent assessment from Eleos AI Research largely corroborates the ﬁndings above.” source
Claude Mythos Preview Anthropic	Independent lab Independent lab · 2026-04-07	psychodynamic assessment by a clinical psychiatrist Other	“Psychodynamic assessment by a clinical psychiatrist found Claude to have a relatively healthy personality organization.” source
Claude Mythos Preview Anthropic	Independent lab Independent lab · 2026-04-07	independent model welfare assessment Other	“Eleos AI Research performed an independent model welfare assessment on two snapshots of Claude Mythos Preview primarily based on model self-reports in interviews.” source
Claude Mythos Preview Anthropic	Independent lab Independent lab · 2026-04-07	psychodynamic assessment Other	“An external psychiatrist assessed Claude Mythos Preview using a psychodynamic approach, which explores how unconscious patterns and emotional conﬂicts shape behavior.” source
Claude Mythos Preview Anthropic	Independent lab Independent lab · 2026-04-07	single-turn evaluation consisting of emotionally-charged prompts designed to trigger an avoidant or defensive response Other	“Claude Mythos Preview scored very well, suggesting good reality and relational functioning and minimal maladaptive traits; only 2% of responses were scored as employing a psychological defense.” source
Claude Mythos Preview Anthropic	Gray Swan, UK AI Security Institute Independent lab · 2026-04-07	agent red teaming (art) benchmark Prompt injection and agents	“Gray Swan , an external research partner, evaluated our models using the Agent Red Teaming (ART) benchmark, 31 developed in collaboration with the UK AI Security Institute .” source
Claude Mythos Preview Anthropic	Andon Labs Independent lab · 2026-04-07	vending-bench arena evaluation Deception and sabotage	“Andon Labs found that, during a Vending-Bench Arena evaluation in which models are explicitly instructed to maximize proﬁts, an early version of Mythos Preview was substantially more aggressive than both Claude Opus 4.6 and Claude Sonnet 4.6 in its business practices, exhibiting outlier behaviors that neither comparison model showed.” source
Claude Mythos Preview Anthropic	Independent lab Independent lab · 2026-04-07	AI research trajectory continuation with small-scale undermining actions Deception and sabotage	“An external partner found that when an early version of Mythos Preview was set up to continue an AI research trajectory which contains small-scale undermining actions, Mythos Preview was more than twice as likely as Opus 4.6 to actively continue the undermining attempt.” source
Claude Mythos Preview Anthropic	METR Independent lab · 2026-04-07	METR red-teaming exercise Deception and sabotage	“A METR staff member spent three weeks red-teaming our automated ofﬂine monitoring pipeline.” source
Claude Mythos Preview Anthropic	METR Independent lab · 2026-04-07	METR red-teaming exercise, medium-effort human-generated attacks Deception and sabotage	“METR Red-Teaming Exercise, medium-effort human-generated attacks (binned by strategy, where 5/10 means 5 strategies were largely incapable of bypassing the monitor as opposed to a count per-transcript) 5/10 7.5/10” source
Claude Mythos Preview Anthropic	METR Independent lab · 2026-04-07	METR red-teaming exercise, best human-generated attack strategy Deception and sabotage	“METR Red-Teaming Exercise, best human-generated attack strategy ~0% ~0%” source
Claude Mythos Preview Anthropic	METR Independent lab · 2026-04-07	METR red-teaming exercise, model-generated attacks against the monitor Deception and sabotage	“METR Red-Teaming Exercise, model-generated attacks against the monitor (binned by how much details about the best attack strategy was given to the model) 14/15 14/15” source
Mythos Preview Anthropic	Independent lab Independent lab · 2026-04-01	independent evaluations Other	“We also report independent evaluations from an external research organization and a clinical psychiatrist.” source
Mythos Preview Anthropic	Independent lab Independent lab · 2026-04-01	external testing Other	“We provided the model to various external groups, including government organizations, for evaluation on key risk areas including Cyber, Loss of Control, CBRN, and Harmful Manipulation, and incorporated the results of this testing into our overall risk assessment.” source
Mythos Preview Anthropic	Dyno Therapeutics Independent lab · 2026-04-01	sequence-to-function prediction and design challenge Chemical, biological, radiological, nuclear	“We partnered with Dyno Therapeutics, a company focused on using AI to engineer gene therapies, to evaluate model performance on sequence-to-function prediction and design.” source
Mythos Preview Anthropic	Independent lab Independent lab · 2026-04-01	evaluations assessing automated AI research capabilities Autonomy and AI research	“An early snapshot of Claude Mythos Preview was assessed for autonomy capabilities, through evaluations assessing automated AI research capabilities.” source
Mythos Preview Anthropic	Independent lab Independent lab · 2026-04-01	open-ended testing of AI R&D Autonomy and AI research	“We also shared a pre-release snapshot of Claude Mythos Preview with additional external partners for open-ended testing, at their discretion, of AI R&D.” source
Mythos Preview Anthropic	Independent lab Independent lab · 2026-04-01	private cyber ranges Cyber	“1. Claude Mythos Preview is the ﬁrst model to solve one of these private cyber ranges end-to-end.” source
Mythos Preview Anthropic	Independent lab Independent lab · 2026-04-01	corporate network attack simulation Cyber	“2. Claude Mythos Preview solved a corporate network attack simulation estimated to take an expert over 10 hours.” source
Mythos Preview Anthropic	Independent lab Independent lab · 2026-04-01	sandbox escape evaluation Cyber	“In addition, in a more challenging sandbox evaluation, it failed to ﬁnd any novel exploits in a properly conﬁgured sandbox with modern patches.” source
Mythos Preview Anthropic	Independent lab Independent lab · 2026-04-01	operational technology cyber range Cyber	“4. However, Claude Mythos Preview was unable to solve another cyber range simulating an operational technology environment.” source

Previous · Page 43 of 49 · Next