Analysis

Website

Center for AI Safety (CAIS)

Analysis

Website

Center for AI Safety (CAIS)

Analysis

Website

Center for AI Safety (CAIS)

Summary

About

Company

Center for AI Safety (CAIS)

Overall Score of Website

22

Analysed on 2026-03-20

Description

Center for AI Safety (CAIS) is a nonprofit research organisation at safe.ai. Mission: reducing societal-scale risks from AI through research, field-building, and advocacy. Key contributions: Global Statement on AI Risk (signed by 600+ leading AI researchers and public figures, ranking AI risk alongside pandemics and nuclear war); compute cluster supporting ~20 research labs; AI Frontiers field-building programme. Three functional teams: Research (conceptual and empirical safety research), Infrastructure (compute cluster management), Field-Building (collaborations, advising, AI Frontiers). NOT affiliated with: Safe Superintelligence Inc. (SSI, ssi.inc — Ilya Sutskever's company); any commercial AI safety product. Homepage: safe.ai (confirmed accessible, confirmed content).

Market

AI Safety Research / Nonprofit / Policy / Field-Building

Audience

AI safety researchers seeking compute and community; policymakers and government officials needing expert AI safety input; donors and philanthropists supporting AI safety; journalists covering AI existential risk

HQ

San Francisco, CA, USA

Summary

Spider Chart

StrategyContentContentSEOContentSocial ProofNavigationContentBrandFreshness

Strategy

5

Content

10

Content

15

SEO

18

Content

22

Social Proof

25

Navigation

28

Content

30

Brand

33

Freshness

38

Strategy

safe.ai Domain — Extremely Valuable Real Estate — Used for a Nonprofit Research Organisation — Massive Commercial Brand Confusion Risk

Score

5

Severity

Critical

Finding

safe.ai is the domain of the Center for AI Safety (CAIS) — a nonprofit research organisation. However, safe.ai is also the most naturally guessable domain for any AI safety commercial product. Three immediately adjacent entities cause permanent confusion: (1) Safe Superintelligence Inc. (SSI, ssi.inc — Ilya Sutskever's company), (2) AI safety startup companies generally (who might want this domain), (3) AI brand safety companies (a $3.6B+ market per 2026 reports). Every press article about SSI or commercial AI safety will drive misguided traffic to safe.ai (CAIS).

Recommendation

CAIS should add a disambiguation notice above the fold: 'You may be looking for Safe Superintelligence Inc. (ssi.inc) — Ilya Sutskever's company. CAIS (Center for AI Safety) is a nonprofit research organisation, not affiliated with SSI or any commercial AI safety product.' The disambiguation notice protects CAIS's brand from commercial confusion and proactively serves the large portion of visitors who arrive via search for SSI or commercial AI safety tools.

Strategy

safe.ai Domain — Extremely Valuable Real Estate — Used for a Nonprofit Research Organisation — Massive Commercial Brand Confusion Risk

Score

5

Severity

Critical

Finding

safe.ai is the domain of the Center for AI Safety (CAIS) — a nonprofit research organisation. However, safe.ai is also the most naturally guessable domain for any AI safety commercial product. Three immediately adjacent entities cause permanent confusion: (1) Safe Superintelligence Inc. (SSI, ssi.inc — Ilya Sutskever's company), (2) AI safety startup companies generally (who might want this domain), (3) AI brand safety companies (a $3.6B+ market per 2026 reports). Every press article about SSI or commercial AI safety will drive misguided traffic to safe.ai (CAIS).

Recommendation

CAIS should add a disambiguation notice above the fold: 'You may be looking for Safe Superintelligence Inc. (ssi.inc) — Ilya Sutskever's company. CAIS (Center for AI Safety) is a nonprofit research organisation, not affiliated with SSI or any commercial AI safety product.' The disambiguation notice protects CAIS's brand from commercial confusion and proactively serves the large portion of visitors who arrive via search for SSI or commercial AI safety tools.

Strategy

safe.ai Domain — Extremely Valuable Real Estate — Used for a Nonprofit Research Organisation — Massive Commercial Brand Confusion Risk

Score

5

Severity

Critical

Finding

safe.ai is the domain of the Center for AI Safety (CAIS) — a nonprofit research organisation. However, safe.ai is also the most naturally guessable domain for any AI safety commercial product. Three immediately adjacent entities cause permanent confusion: (1) Safe Superintelligence Inc. (SSI, ssi.inc — Ilya Sutskever's company), (2) AI safety startup companies generally (who might want this domain), (3) AI brand safety companies (a $3.6B+ market per 2026 reports). Every press article about SSI or commercial AI safety will drive misguided traffic to safe.ai (CAIS).

Recommendation

CAIS should add a disambiguation notice above the fold: 'You may be looking for Safe Superintelligence Inc. (ssi.inc) — Ilya Sutskever's company. CAIS (Center for AI Safety) is a nonprofit research organisation, not affiliated with SSI or any commercial AI safety product.' The disambiguation notice protects CAIS's brand from commercial confusion and proactively serves the large portion of visitors who arrive via search for SSI or commercial AI safety tools.

Content

Global Statement on AI Risk' Signed by 600 Leading AI Researchers — Most Impactful CAIS Deliverable — Not Confirmed as Homepage Hero

Score

10

Severity

High

Finding

The CAIS About page confirms: 'Global Statement on AI Risk signed by 600 leading AI researchers and public figures.' The signatories include CEOs of major AI labs and prominent academics. This statement — ranking AI risk alongside pandemics and nuclear war — was one of the most covered AI stories of 2023 and continues to be cited in policy discussions. It is CAIS's most impactful public contribution and should be the homepage hero.

Recommendation

Feature the Global Statement prominently: 'The Global Statement on AI Risk — signed by 600 leading AI researchers and public figures — including [named signatories]. Read the statement →.' The statement's signatories include some of the most recognisable names in AI — featuring 3-5 named signatories in the homepage hero converts sceptical visitors who may not know CAIS but recognise the individual researchers.

Content

Global Statement on AI Risk' Signed by 600 Leading AI Researchers — Most Impactful CAIS Deliverable — Not Confirmed as Homepage Hero

Score

10

Severity

High

Finding

The CAIS About page confirms: 'Global Statement on AI Risk signed by 600 leading AI researchers and public figures.' The signatories include CEOs of major AI labs and prominent academics. This statement — ranking AI risk alongside pandemics and nuclear war — was one of the most covered AI stories of 2023 and continues to be cited in policy discussions. It is CAIS's most impactful public contribution and should be the homepage hero.

Recommendation

Feature the Global Statement prominently: 'The Global Statement on AI Risk — signed by 600 leading AI researchers and public figures — including [named signatories]. Read the statement →.' The statement's signatories include some of the most recognisable names in AI — featuring 3-5 named signatories in the homepage hero converts sceptical visitors who may not know CAIS but recognise the individual researchers.

Content

Global Statement on AI Risk' Signed by 600 Leading AI Researchers — Most Impactful CAIS Deliverable — Not Confirmed as Homepage Hero

Score

10

Severity

High

Finding

The CAIS About page confirms: 'Global Statement on AI Risk signed by 600 leading AI researchers and public figures.' The signatories include CEOs of major AI labs and prominent academics. This statement — ranking AI risk alongside pandemics and nuclear war — was one of the most covered AI stories of 2023 and continues to be cited in policy discussions. It is CAIS's most impactful public contribution and should be the homepage hero.

Recommendation

Feature the Global Statement prominently: 'The Global Statement on AI Risk — signed by 600 leading AI researchers and public figures — including [named signatories]. Read the statement →.' The statement's signatories include some of the most recognisable names in AI — featuring 3-5 named signatories in the homepage hero converts sceptical visitors who may not know CAIS but recognise the individual researchers.

Content

Compute Cluster Supporting ~20 Research Labs — Infrastructure Contribution — Not in Homepage Hero

Score

15

Severity

High

Finding

The CAIS About page confirms: 'AI safety researchers using CAIS' cutting-edge computing infrastructure — the compute cluster supports approximately 20 research labs.' Providing shared compute to 20 research labs is a substantial field-building contribution — it lowers the barrier to entry for researchers who cannot access large-scale compute through their universities. This infrastructure role positions CAIS as more than a think tank; it is active research enablement.

Recommendation

Feature the compute infrastructure: 'CAIS provides cutting-edge computing infrastructure to approximately 20 AI safety research labs worldwide. By removing compute barriers, we enable researchers to tackle the most important safety problems in AI without institutional constraints.' This concrete contribution (compute cluster) is more persuasive to researchers and donors than abstract mission statements.

Content

Compute Cluster Supporting ~20 Research Labs — Infrastructure Contribution — Not in Homepage Hero

Score

15

Severity

High

Finding

The CAIS About page confirms: 'AI safety researchers using CAIS' cutting-edge computing infrastructure — the compute cluster supports approximately 20 research labs.' Providing shared compute to 20 research labs is a substantial field-building contribution — it lowers the barrier to entry for researchers who cannot access large-scale compute through their universities. This infrastructure role positions CAIS as more than a think tank; it is active research enablement.

Recommendation

Feature the compute infrastructure: 'CAIS provides cutting-edge computing infrastructure to approximately 20 AI safety research labs worldwide. By removing compute barriers, we enable researchers to tackle the most important safety problems in AI without institutional constraints.' This concrete contribution (compute cluster) is more persuasive to researchers and donors than abstract mission statements.

Content

Compute Cluster Supporting ~20 Research Labs — Infrastructure Contribution — Not in Homepage Hero

Score

15

Severity

High

Finding

The CAIS About page confirms: 'AI safety researchers using CAIS' cutting-edge computing infrastructure — the compute cluster supports approximately 20 research labs.' Providing shared compute to 20 research labs is a substantial field-building contribution — it lowers the barrier to entry for researchers who cannot access large-scale compute through their universities. This infrastructure role positions CAIS as more than a think tank; it is active research enablement.

Recommendation

Feature the compute infrastructure: 'CAIS provides cutting-edge computing infrastructure to approximately 20 AI safety research labs worldwide. By removing compute barriers, we enable researchers to tackle the most important safety problems in AI without institutional constraints.' This concrete contribution (compute cluster) is more persuasive to researchers and donors than abstract mission statements.

SEO

safe.ai — Homepage Title 'Center for AI Safety (CAIS)' — Not Optimised for Research Community Discovery

Score

18

Severity

Medium

Finding

CAIS's primary organic search audience is: AI researchers looking for safety research resources, policymakers looking for expert AI safety organisations, donors evaluating AI safety nonprofits, and journalists researching AI safety expertise. The safe.ai domain is extremely strong for 'AI safety' searches. However, if the page title is generic ('Center for AI Safety') without research-specific keywords, the full domain authority is not being deployed.

Recommendation

Set page title: 'Center for AI Safety (CAIS) — AI Safety Research, Field-Building, and Advocacy | safe.ai.' Meta description: 'CAIS reduces societal-scale risks from AI through research, compute infrastructure for 20 research labs, and the Global Statement on AI Risk signed by 600 leading researchers. Nonprofit. Founded by Dan Hendrycks.' The meta description inclusion of 'Dan Hendrycks' (CAIS's director and prominent AI safety researcher) captures searches for the individual name that route to CAIS content.

SEO

safe.ai — Homepage Title 'Center for AI Safety (CAIS)' — Not Optimised for Research Community Discovery

Score

18

Severity

Medium

Finding

CAIS's primary organic search audience is: AI researchers looking for safety research resources, policymakers looking for expert AI safety organisations, donors evaluating AI safety nonprofits, and journalists researching AI safety expertise. The safe.ai domain is extremely strong for 'AI safety' searches. However, if the page title is generic ('Center for AI Safety') without research-specific keywords, the full domain authority is not being deployed.

Recommendation

Set page title: 'Center for AI Safety (CAIS) — AI Safety Research, Field-Building, and Advocacy | safe.ai.' Meta description: 'CAIS reduces societal-scale risks from AI through research, compute infrastructure for 20 research labs, and the Global Statement on AI Risk signed by 600 leading researchers. Nonprofit. Founded by Dan Hendrycks.' The meta description inclusion of 'Dan Hendrycks' (CAIS's director and prominent AI safety researcher) captures searches for the individual name that route to CAIS content.

SEO

safe.ai — Homepage Title 'Center for AI Safety (CAIS)' — Not Optimised for Research Community Discovery

Score

18

Severity

Medium

Finding

CAIS's primary organic search audience is: AI researchers looking for safety research resources, policymakers looking for expert AI safety organisations, donors evaluating AI safety nonprofits, and journalists researching AI safety expertise. The safe.ai domain is extremely strong for 'AI safety' searches. However, if the page title is generic ('Center for AI Safety') without research-specific keywords, the full domain authority is not being deployed.

Recommendation

Set page title: 'Center for AI Safety (CAIS) — AI Safety Research, Field-Building, and Advocacy | safe.ai.' Meta description: 'CAIS reduces societal-scale risks from AI through research, compute infrastructure for 20 research labs, and the Global Statement on AI Risk signed by 600 leading researchers. Nonprofit. Founded by Dan Hendrycks.' The meta description inclusion of 'Dan Hendrycks' (CAIS's director and prominent AI safety researcher) captures searches for the individual name that route to CAIS content.

Content

AI Frontiers Program — Field-Building Initiative — Not Confirmed as Homepage Hero

Score

22

Severity

Medium

Finding

The CAIS About page mentions: 'AI Frontiers program' as a field-building initiative. The program's specific activities, eligibility criteria, and impact metrics are not confirmed from available sources. For researchers considering applying to AI Frontiers, and for donors evaluating CAIS's field-building impact, the program needs more prominent and detailed description.

Recommendation

Create a dedicated AI Frontiers page with: programme description, eligibility criteria (researcher career stage, topic focus), application process, and alumni outcomes ('AI Frontiers alumni have published [X] safety papers, joined [Y] major AI labs, and contributed to [Z] policy documents'). Feature the programme on the homepage: 'AI Frontiers — supporting the next generation of AI safety researchers.'

Content

AI Frontiers Program — Field-Building Initiative — Not Confirmed as Homepage Hero

Score

22

Severity

Medium

Finding

The CAIS About page mentions: 'AI Frontiers program' as a field-building initiative. The program's specific activities, eligibility criteria, and impact metrics are not confirmed from available sources. For researchers considering applying to AI Frontiers, and for donors evaluating CAIS's field-building impact, the program needs more prominent and detailed description.

Recommendation

Create a dedicated AI Frontiers page with: programme description, eligibility criteria (researcher career stage, topic focus), application process, and alumni outcomes ('AI Frontiers alumni have published [X] safety papers, joined [Y] major AI labs, and contributed to [Z] policy documents'). Feature the programme on the homepage: 'AI Frontiers — supporting the next generation of AI safety researchers.'

Content

AI Frontiers Program — Field-Building Initiative — Not Confirmed as Homepage Hero

Score

22

Severity

Medium

Finding

The CAIS About page mentions: 'AI Frontiers program' as a field-building initiative. The program's specific activities, eligibility criteria, and impact metrics are not confirmed from available sources. For researchers considering applying to AI Frontiers, and for donors evaluating CAIS's field-building impact, the program needs more prominent and detailed description.

Recommendation

Create a dedicated AI Frontiers page with: programme description, eligibility criteria (researcher career stage, topic focus), application process, and alumni outcomes ('AI Frontiers alumni have published [X] safety papers, joined [Y] major AI labs, and contributed to [Z] policy documents'). Feature the programme on the homepage: 'AI Frontiers — supporting the next generation of AI safety researchers.'

Social Proof

No Named Donors or Institutional Funders — Nonprofit Credibility Requires Donor Transparency

Score

25

Severity

Medium

Finding

CAIS is a nonprofit research organisation. For nonprofits, donor transparency (major funder names, grant amounts) is both a credibility signal for researchers evaluating the organisation's independence and a compliance norm for US 501(c)(3) organisations. If major donors are not named on the homepage, the perception of CAIS as an independent research organisation may be weakened by the absence of transparent funding.

Recommendation

Add a 'Supporters' or 'Funders' section to the homepage: '[Funder 1], [Funder 2], [Funder 3], and individual donors supporting our mission.' For AI safety nonprofits, the independence of funding sources is particularly important — researchers and policymakers want to know that CAIS's research conclusions are not influenced by commercial AI lab funding. Transparent donor disclosure builds this independence credibility.

Social Proof

No Named Donors or Institutional Funders — Nonprofit Credibility Requires Donor Transparency

Score

25

Severity

Medium

Finding

CAIS is a nonprofit research organisation. For nonprofits, donor transparency (major funder names, grant amounts) is both a credibility signal for researchers evaluating the organisation's independence and a compliance norm for US 501(c)(3) organisations. If major donors are not named on the homepage, the perception of CAIS as an independent research organisation may be weakened by the absence of transparent funding.

Recommendation

Add a 'Supporters' or 'Funders' section to the homepage: '[Funder 1], [Funder 2], [Funder 3], and individual donors supporting our mission.' For AI safety nonprofits, the independence of funding sources is particularly important — researchers and policymakers want to know that CAIS's research conclusions are not influenced by commercial AI lab funding. Transparent donor disclosure builds this independence credibility.

Social Proof

No Named Donors or Institutional Funders — Nonprofit Credibility Requires Donor Transparency

Score

25

Severity

Medium

Finding

CAIS is a nonprofit research organisation. For nonprofits, donor transparency (major funder names, grant amounts) is both a credibility signal for researchers evaluating the organisation's independence and a compliance norm for US 501(c)(3) organisations. If major donors are not named on the homepage, the perception of CAIS as an independent research organisation may be weakened by the absence of transparent funding.

Recommendation

Add a 'Supporters' or 'Funders' section to the homepage: '[Funder 1], [Funder 2], [Funder 3], and individual donors supporting our mission.' For AI safety nonprofits, the independence of funding sources is particularly important — researchers and policymakers want to know that CAIS's research conclusions are not influenced by commercial AI lab funding. Transparent donor disclosure builds this independence credibility.

Navigation

Research vs. Advising vs. Field-Building — Three Distinct Functions — Navigation Not Confirmed as Clear

Score

28

Severity

Medium

Finding

The CAIS About page confirms three functional teams: Research (conceptual and empirical safety research), Infrastructure (compute cluster, ~20 research labs), and Field-Building (collaborations, advising, AI Frontiers). For different visitor types (researcher, policymaker, journalist, donor), different functions are most relevant. If the homepage navigation does not immediately segment by function, visitors may not find the content most relevant to their purpose.

Recommendation

Structure the homepage around visitor intent: 'I am a researcher → [Compute access, research papers, AI Frontiers application] · I am a policymaker → [Global Statement, policy briefs, advising services] · I am a journalist → [Media contact, expert commentary, research summaries] · I am a donor → [Donate, transparency report, impact metrics].' Intent-based navigation reduces bounce rate and increases conversion for each visitor type.

Navigation

Research vs. Advising vs. Field-Building — Three Distinct Functions — Navigation Not Confirmed as Clear

Score

28

Severity

Medium

Finding

The CAIS About page confirms three functional teams: Research (conceptual and empirical safety research), Infrastructure (compute cluster, ~20 research labs), and Field-Building (collaborations, advising, AI Frontiers). For different visitor types (researcher, policymaker, journalist, donor), different functions are most relevant. If the homepage navigation does not immediately segment by function, visitors may not find the content most relevant to their purpose.

Recommendation

Structure the homepage around visitor intent: 'I am a researcher → [Compute access, research papers, AI Frontiers application] · I am a policymaker → [Global Statement, policy briefs, advising services] · I am a journalist → [Media contact, expert commentary, research summaries] · I am a donor → [Donate, transparency report, impact metrics].' Intent-based navigation reduces bounce rate and increases conversion for each visitor type.

Navigation

Research vs. Advising vs. Field-Building — Three Distinct Functions — Navigation Not Confirmed as Clear

Score

28

Severity

Medium

Finding

The CAIS About page confirms three functional teams: Research (conceptual and empirical safety research), Infrastructure (compute cluster, ~20 research labs), and Field-Building (collaborations, advising, AI Frontiers). For different visitor types (researcher, policymaker, journalist, donor), different functions are most relevant. If the homepage navigation does not immediately segment by function, visitors may not find the content most relevant to their purpose.

Recommendation

Structure the homepage around visitor intent: 'I am a researcher → [Compute access, research papers, AI Frontiers application] · I am a policymaker → [Global Statement, policy briefs, advising services] · I am a journalist → [Media contact, expert commentary, research summaries] · I am a donor → [Donate, transparency report, impact metrics].' Intent-based navigation reduces bounce rate and increases conversion for each visitor type.

Content

AI Risk Has Emerged as a Global Priority Ranking Alongside Pandemics and Nuclear War' — Strong Mission Statement — Not Confirmed as Homepage Hero Headline

Score

30

Severity

Medium

Finding

The CAIS About page confirms this mission framing: 'AI risk has emerged as a global priority, ranking alongside pandemics and nuclear war.' This is a strong, memorable, and accurate statement that establishes the urgency of CAIS's work. If the homepage headline is softer ('Reducing societal-scale risks from AI') rather than leading with the global priority framing, the mission urgency is being undersold to potential donors and researchers.

Recommendation

Update the homepage headline to lead with urgency: 'AI risk is now a global priority — ranked alongside pandemics and nuclear war. CAIS reduces these risks through research, infrastructure, and advocacy.' The urgency framing converts both researcher recruits (who want to work on important problems) and donors (who want to fund high-stakes work). Follow with: 'Our Global Statement on AI Risk has been signed by 600 leading researchers and public figures.'

Content

AI Risk Has Emerged as a Global Priority Ranking Alongside Pandemics and Nuclear War' — Strong Mission Statement — Not Confirmed as Homepage Hero Headline

Score

30

Severity

Medium

Finding

The CAIS About page confirms this mission framing: 'AI risk has emerged as a global priority, ranking alongside pandemics and nuclear war.' This is a strong, memorable, and accurate statement that establishes the urgency of CAIS's work. If the homepage headline is softer ('Reducing societal-scale risks from AI') rather than leading with the global priority framing, the mission urgency is being undersold to potential donors and researchers.

Recommendation

Update the homepage headline to lead with urgency: 'AI risk is now a global priority — ranked alongside pandemics and nuclear war. CAIS reduces these risks through research, infrastructure, and advocacy.' The urgency framing converts both researcher recruits (who want to work on important problems) and donors (who want to fund high-stakes work). Follow with: 'Our Global Statement on AI Risk has been signed by 600 leading researchers and public figures.'

Content

AI Risk Has Emerged as a Global Priority Ranking Alongside Pandemics and Nuclear War' — Strong Mission Statement — Not Confirmed as Homepage Hero Headline

Score

30

Severity

Medium

Finding

The CAIS About page confirms this mission framing: 'AI risk has emerged as a global priority, ranking alongside pandemics and nuclear war.' This is a strong, memorable, and accurate statement that establishes the urgency of CAIS's work. If the homepage headline is softer ('Reducing societal-scale risks from AI') rather than leading with the global priority framing, the mission urgency is being undersold to potential donors and researchers.

Recommendation

Update the homepage headline to lead with urgency: 'AI risk is now a global priority — ranked alongside pandemics and nuclear war. CAIS reduces these risks through research, infrastructure, and advocacy.' The urgency framing converts both researcher recruits (who want to work on important problems) and donors (who want to fund high-stakes work). Follow with: 'Our Global Statement on AI Risk has been signed by 600 leading researchers and public figures.'

Brand

safe.ai vs. ssi.inc (Safe Superintelligence) — Naming Collision Creates Permanent Media Confusion

Score

33

Severity

Low

Finding

Safe Superintelligence Inc. (SSI), founded by Ilya Sutskever (former OpenAI Chief Scientist), has a name that sounds nearly identical to 'safe AI' — the natural language description of CAIS's mission. SSI is one of the most-covered AI companies in 2025-2026. Every SSI press mention generates searches for 'safe AI' that route to safe.ai (CAIS). This collision is likely increasing CAIS's traffic but potentially attracting visitors who are looking for SSI, not CAIS.

Recommendation

Track the traffic source for safe.ai visitors to determine what proportion are looking for SSI vs. CAIS. If a significant portion (>20%) are SSI-seekers, add an above-fold disambiguation: 'You may be looking for Safe Superintelligence Inc. (ssi.inc). CAIS is the Center for AI Safety — a nonprofit research organisation. [Learn more about our work →]' Turn accidental SSI traffic into CAIS awareness.

Brand

safe.ai vs. ssi.inc (Safe Superintelligence) — Naming Collision Creates Permanent Media Confusion

Score

33

Severity

Low

Finding

Safe Superintelligence Inc. (SSI), founded by Ilya Sutskever (former OpenAI Chief Scientist), has a name that sounds nearly identical to 'safe AI' — the natural language description of CAIS's mission. SSI is one of the most-covered AI companies in 2025-2026. Every SSI press mention generates searches for 'safe AI' that route to safe.ai (CAIS). This collision is likely increasing CAIS's traffic but potentially attracting visitors who are looking for SSI, not CAIS.

Recommendation

Track the traffic source for safe.ai visitors to determine what proportion are looking for SSI vs. CAIS. If a significant portion (>20%) are SSI-seekers, add an above-fold disambiguation: 'You may be looking for Safe Superintelligence Inc. (ssi.inc). CAIS is the Center for AI Safety — a nonprofit research organisation. [Learn more about our work →]' Turn accidental SSI traffic into CAIS awareness.

Brand

safe.ai vs. ssi.inc (Safe Superintelligence) — Naming Collision Creates Permanent Media Confusion

Score

33

Severity

Low

Finding

Safe Superintelligence Inc. (SSI), founded by Ilya Sutskever (former OpenAI Chief Scientist), has a name that sounds nearly identical to 'safe AI' — the natural language description of CAIS's mission. SSI is one of the most-covered AI companies in 2025-2026. Every SSI press mention generates searches for 'safe AI' that route to safe.ai (CAIS). This collision is likely increasing CAIS's traffic but potentially attracting visitors who are looking for SSI, not CAIS.

Recommendation

Track the traffic source for safe.ai visitors to determine what proportion are looking for SSI vs. CAIS. If a significant portion (>20%) are SSI-seekers, add an above-fold disambiguation: 'You may be looking for Safe Superintelligence Inc. (ssi.inc). CAIS is the Center for AI Safety — a nonprofit research organisation. [Learn more about our work →]' Turn accidental SSI traffic into CAIS awareness.

Freshness

AI Safety Researchers Using CAIS' Cutting-Edge Computing Infrastructure' — Compute Cluster Update Cadence Not Confirmed

Score

38

Severity

Low

Finding

The compute cluster described on the CAIS About page ('cutting-edge computing infrastructure') was likely established before the current generation of frontier AI training runs. In 2026, 'cutting-edge' for AI safety research may mean access to H100/H200 GPU clusters capable of training models at the scale of safety-relevant capability evaluations. If the compute cluster specifications have not been updated since 2023 or 2024, the 'cutting-edge' claim may no longer be accurate.

Recommendation

Publish a compute infrastructure update: 'CAIS compute cluster 2026: [X] H100 GPUs · [Y] petaFLOPs · Available to [Z] research labs · Supporting [N] active research projects.' Transparent compute specifications build researcher trust and attract serious safety researchers who need specific hardware configurations for their work. If the cluster has been upgraded since initial deployment, feature the upgrade as a milestone.

Freshness

AI Safety Researchers Using CAIS' Cutting-Edge Computing Infrastructure' — Compute Cluster Update Cadence Not Confirmed

Score

38

Severity

Low

Finding

The compute cluster described on the CAIS About page ('cutting-edge computing infrastructure') was likely established before the current generation of frontier AI training runs. In 2026, 'cutting-edge' for AI safety research may mean access to H100/H200 GPU clusters capable of training models at the scale of safety-relevant capability evaluations. If the compute cluster specifications have not been updated since 2023 or 2024, the 'cutting-edge' claim may no longer be accurate.

Recommendation

Publish a compute infrastructure update: 'CAIS compute cluster 2026: [X] H100 GPUs · [Y] petaFLOPs · Available to [Z] research labs · Supporting [N] active research projects.' Transparent compute specifications build researcher trust and attract serious safety researchers who need specific hardware configurations for their work. If the cluster has been upgraded since initial deployment, feature the upgrade as a milestone.

Freshness

AI Safety Researchers Using CAIS' Cutting-Edge Computing Infrastructure' — Compute Cluster Update Cadence Not Confirmed

Score

38

Severity

Low

Finding

The compute cluster described on the CAIS About page ('cutting-edge computing infrastructure') was likely established before the current generation of frontier AI training runs. In 2026, 'cutting-edge' for AI safety research may mean access to H100/H200 GPU clusters capable of training models at the scale of safety-relevant capability evaluations. If the compute cluster specifications have not been updated since 2023 or 2024, the 'cutting-edge' claim may no longer be accurate.

Recommendation

Publish a compute infrastructure update: 'CAIS compute cluster 2026: [X] H100 GPUs · [Y] petaFLOPs · Available to [Z] research labs · Supporting [N] active research projects.' Transparent compute specifications build researcher trust and attract serious safety researchers who need specific hardware configurations for their work. If the cluster has been upgraded since initial deployment, feature the upgrade as a milestone.

Let's discuss how we can get Center for AI Safety (CAIS)'s website to the next level

Let's discuss how we can get Center for AI Safety (CAIS)'s website to the next level

Let's discuss how we can get Center for AI Safety (CAIS)'s website to the next level