safe.ai

The Center for AI Safety (CAIS, pronounced "case") is a San Francisco-based research and field-building nonprofit whose mission is to reduce societal-scale risks associated with AI. Its work falls into three categories: conducting safety research, building the field of AI safety researchers, and advocating for safety standards. Below is an overview of the specific projects and resources it offers.

Three categories of work

CAIS describes its mission as reducing societal-scale risks from AI through three complementary activities:

• Safety research — technical and conceptual work on unsolved problems in AI safety.
• Field-building — training and supporting people who will work on AI safety.
• Advocacy — promoting safety standards.

The organization notes that preventing extreme risks from AI requires more than technical work alone, which is why it takes a multidisciplinary approach across academic disciplines, public and private entities, and the general public.

Research and field-building projects

ML Safety Infrastructure

A featured project under CAIS's field-building work, aimed at supporting the technical infrastructure the AI safety research community relies on.

AI and Society Fellowship

A three-month research program that investigates the societal impacts of advanced AI and the institutions and policies that could help societies respond well. According to the site, applications close March 24.

Philosophy Fellowship

A fellowship program listed among CAIS's field-building projects.

AI Safety, Ethics, & Society Training

A course described as a comprehensive introduction to how current AI systems work, their societal-scale risks, and how to manage them. CAIS states it has trained 1000+ future AI safety leaders through this program.

CAIS Compute Cluster

A compute cluster made available as part of CAIS's work, intended to support AI safety research.

Introduction to AI Safety (with Coursera)

A course offered in partnership with Coursera. It covers the drivers of modern AI development, major risks from AI — including malicious use, competitive pressures, and loss of control — and governance strategies across different scales for mitigating these risks.

Resources and publications

CAIS publishes and maintains several public resources:

The site also includes a "Statement on AI Extinction Risk" and a Frequently Asked Questions section, and organizes its content under Work, About, and Resources sections.

How to engage

If you want to go deeper, the most direct entry points are the Coursera course for a structured introduction, the AI and Society Fellowship for a research program (note the March 24 application deadline), and the blog, newsletter, and impact reports for ongoing updates on CAIS's work. The site also lists Careers and Donate pages for those interested in supporting the organization directly.

The Center for AI Safety (CAIS) focuses on societal-scale risks from AI—risks large enough to affect whole societies rather than only individual users or companies. CAIS describes AI as a technology that can "profoundly benefit the world" while also carrying "inherent risks, including some which are potentially catastrophic." Its stated mission is to reduce those risks through safety research, field-building, and advocacy for safety standards. This article explains what that risk framing includes and why CAIS treats it as a multidisciplinary problem.

How CAIS frames AI risk

CAIS does not present AI as either purely beneficial or purely dangerous. Its position, as stated on its site, is conditional: AI's potential to benefit the world depends on whether it can be "developed and used safely."

That framing matters because it sets the scope of CAIS's work. The organization is not focused on narrow product bugs or ordinary software failures. It is focused on risks that scale with capability—the kind that become harder to contain as systems become more powerful and more widely deployed.

CAIS also notes a gap between progress and understanding: "In contrast to the dramatic progress in AI, many basic problems in AI safety have yet to be solved." The risk concern, in other words, is not only about what AI can do today, but about the mismatch between rapid capability gains and slower progress on safety.

Categories of risk CAIS highlights

Based on CAIS's own materials, its risk focus spans several categories:

• Malicious use — AI systems used deliberately to cause harm.
• Competitive pressures — dynamics where organizations or countries feel pressured to cut safety corners to keep pace.
• Loss of control — scenarios where humans can no longer reliably direct or constrain advanced systems.
• Catastrophic and extinction-level risk — the upper end of the risk spectrum, which CAIS addresses directly through its Statement on AI Extinction Risk.

CAIS's Coursera course description names malicious use, competitive pressures, and loss of control as "major risks from AI," and pairs them with governance strategies "across different scales." That pairing is notable: CAIS treats technical risk and governance risk as connected rather than separate problems.

Current systems vs. future systems

CAIS's risk page distinguishes between current AI systems and risks that emerge as systems become more advanced. This distinction is useful for readers trying to place CAIS in the broader AI safety landscape:

Why CAIS treats this as more than a technical problem

CAIS Director Dan Hendrycks states: "Preventing extreme risks from AI requires more than just technical work, so CAIS takes a multidisciplinary approach working across academic disciplines, public and private entities, and with the general public."

This is the clearest statement of why CAIS's risk focus looks broader than a typical research lab's. If extreme risk can arise from competitive pressure, misuse, or governance failure, then safety research alone cannot address it. That logic explains CAIS's three-part mission:

1. Conduct safety research — technical and conceptual work on AI safety problems.
2. Build the field — training and supporting people who work on AI safety.
3. Advocate for safety standards — pushing for norms and rules that reduce risk.

The field-building side is visible in programs like the AI Safety, Ethics, & Society training, which CAIS says has reached "1000+ future AI safety leaders," and the AI and Society Fellowship, a three-month research program on the societal impacts of advanced AI and the institutions and policies that could help societies respond.

What "societal-scale" means in practice

The phrase "societal-scale risks" is doing real work in CAIS's framing. It signals risks whose effects are not contained within a single user, company, or transaction—but instead propagate across institutions, economies, or populations.

For a concrete way to think about the difference:

• A biased hiring tool is a serious but bounded harm.
• A system that can be cheaply repurposed for large-scale disinformation, cyberattacks, or bioweapon design assistance is a societal-scale concern.
• A system that resists human oversight at a critical moment is a loss-of-control concern.

CAIS's risk categories map onto the second and third examples more than the first. That is why its materials emphasize catastrophic potential rather than only near-term product harms.

How to use this if you're evaluating CAIS

If you are trying to decide whether CAIS's risk focus matches your own interests, a few practical checks help:

• If you care about near-term, applied AI harms (bias, privacy, misinformation in specific products), CAIS's framing will feel broader and more future-oriented than you may need.
• If you care about catastrophic or existential risk, governance, and field-building, CAIS's mission and programs are directly aligned.
• If you want technical safety research, CAIS lists research as one of its three pillars, alongside field-building and advocacy.
• If you want to get involved, the site points to a fellowship, a training course, a Coursera course, a blog, a newsletter, publications, and impact reports.

The site also provides a Statement on AI Extinction Risk and an FAQ, which are the most direct places to check CAIS's own position on the most severe risk claims rather than relying on secondhand summaries.

The short version

CAIS focuses on AI risks that are large enough to affect society as a whole—malicious use, competitive pressure, loss of control, and catastrophic outcomes—and treats them as problems that require research, a trained field of researchers, and safety standards, not technical fixes alone. Its own framing is conditional rather than alarmist: AI can benefit the world, but only if it is developed and used safely.

The Center for AI Safety (CAIS, pronounced "case") is a San Francisco-based research and field-building nonprofit whose mission is to reduce societal-scale risks associated with AI. It pursues that mission through three routes: conducting safety research, building the field of AI safety researchers, and advocating for safety standards. If you are looking for an organization that treats AI risk as a societal problem rather than only a technical one, CAIS is one of the groups to know.

What CAIS says it does

CAIS describes its work as spanning research, field-building, and advocacy projects. Its stated belief is that AI has the potential to profoundly benefit the world, provided it can be developed and used safely. It also frames its mission against a gap: in contrast to the dramatic progress in AI, many basic problems in AI safety remain unsolved.

The site organizes its work under a few visible areas:

• Research — safety research aimed at societal-scale risks
• Field-building projects — programs that train and connect future AI safety researchers
• Compute cluster — shared compute infrastructure for safety work
• Advocacy — promoting safety standards

The three paths to reducing risk

1. Safety research

CAIS positions research as one leg of the mission, aimed at the basic problems in AI safety that it says are still unsolved. The site also hosts a compute cluster, which supports the idea that some of this work requires infrastructure beyond individual effort.

2. Building the field

This is the most visible part of CAIS's public programming. Examples listed on the site include:

The AI and Society Fellowship page notes applications close March 24, so deadlines are program-specific and worth checking directly if you intend to apply.

3. Advocating for safety standards

The third leg is advocacy — promoting safety standards rather than only producing research. This is consistent with CAIS's public positioning on extreme risk, including its Statement on AI Extinction Risk.

Who runs it

Dan Hendrycks is the Director of the Center for AI Safety and holds a PhD in Computer Science from UC Berkeley. In his words on the site: "Preventing extreme risks from AI requires more than just technical work, so CAIS takes a multidisciplinary approach working across academic disciplines, public and private entities, and with the general public."

That quote is a useful summary of how CAIS distinguishes itself: the approach is explicitly multidisciplinary, not purely technical.

When CAIS is relevant to you

• You want to work in AI safety. The field-building programs — fellowships and courses — are the most direct entry points, and they are aimed at people who may not yet be specialists.
• You want a plain-language framing of AI risk. The Coursera course and the site's "Risks from AI" section cover malicious use, competitive pressures, and loss of control.
• You are looking for an organization's stance on extreme risk. The Statement on AI Extinction Risk is a public artifact you can read and evaluate on its own terms.
• You need compute for safety research. The compute cluster is listed as part of CAIS's work.

What the site does not tell you

The page evidence does not include pricing, membership terms, or login requirements for CAIS programs and resources, so treat cost and access as unknown until you check the specific program page. Program dates and eligibility also vary by offering — the March 24 fellowship deadline is one example, not a general rule.

For anything current — application windows, course availability, or how to get involved — the reliable move is to go to safe.ai and open the specific program page rather than relying on a summary.

The Center for AI Safety (CAIS — pronounced "case") is a San Francisco-based research and field-building nonprofit working to reduce societal-scale risks from AI. You can get involved through four main routes: taking a course, applying to a fellowship, joining the organization through careers, or supporting it through donations and staying informed via its blog and newsletter. Which route fits you depends on whether you want to learn, do research, work, or fund the work.

Learn: Courses and Training

CAIS runs training programs aimed at building the field of AI safety researchers and giving newcomers a structured introduction to the subject.

AI Safety, Ethics, & Society Training

• A course offering a comprehensive introduction to how current AI systems work, their societal-scale risks, and how to manage them.
• CAIS describes it as training "1000+ future AI safety leaders."
• Best suited to people who want a broad grounding across technical, ethical, and societal dimensions rather than a single narrow topic.

Introduction to AI Safety (with Coursera)

• Offered in partnership with Coursera.
• Covers the drivers of modern AI development, major risks from AI — including malicious use, competitive pressures, and loss of control — and governance strategies across different scales for mitigating these risks.
• A reasonable starting point if you prefer a self-paced online format and want the governance and risk framing alongside the technical picture.

If your goal is simply to understand the landscape before committing to research or a career change, start with one of these two courses.

Research: Fellowships and Programs

For people who want to spend focused time on AI safety questions, CAIS lists fellowship programs under its field-building work.

AI and Society Fellowship

• A three-month research program that investigates the societal impacts of advanced AI and the institutions and policies that could help societies respond well.
• Applications close March 24 (as stated on the site).
• Suited to researchers interested in the policy and institutional side of AI risk, not only technical safety.

Philosophy Fellowship

• Listed among CAIS's field-building programs.
• Aimed at those approaching AI safety from a philosophical angle.

Compute Cluster

• CAIS provides a compute cluster as part of its infrastructure for safety research.
• Relevant if your work requires computational resources and you are affiliated with its research efforts.

Before applying, check the current program page for eligibility, dates, and whether applications are open — deadlines and cohort details change.

Work: Careers at CAIS

CAIS lists a Careers page alongside its research, field-building, and advocacy work. If you want to contribute directly rather than through a program, this is the entry point for open roles. The organization describes itself as multidisciplinary, working across academic disciplines, public and private entities, and with the general public — so openings are not limited to technical research.

Support and Stay Informed

If you are not looking to study, research, or work there, you can still engage:

• Donate — CAIS is a nonprofit and lists a Donate page.
• AI Safety Newsletter — a recurring publication for following developments in the field.
• CAIS Blog — posts on the organization's work and AI risk topics.
• Impact Reports — the 2024 and 2023 Impact Reports summarize what the organization has done, useful for judging whether you want to support or join it.

Choosing Your Route

A Note on What CAIS Focuses On

CAIS frames its mission around reducing societal-scale risks from AI through three activities: conducting safety research, building the field of AI safety researchers, and advocating for safety standards. It also published a Statement on AI Extinction Risk. Understanding this framing helps you pick the right way in — the courses and fellowships are the field-building side, while careers and donations support the research and advocacy side.

The Center for AI Safety (CAIS) is one of the organizations behind the public statement that "mitigating the risk of extinction from AI should be a global priority alongside other societal-scale risks such as pandemics and nuclear war." The statement is a short, single-sentence declaration hosted on CAIS's site, and its purpose is to put catastrophic and extinction-level AI risk on the same footing as other widely recognized global threats. It is a position statement, not a technical paper: it asserts priority, not a specific policy or timeline.

What the statement says

The core claim is narrow and deliberately so. It does not argue that extinction is likely, nor does it prescribe how to prevent it. It says only that mitigating the risk of extinction from AI should be treated as a global priority, and it groups that risk with pandemics and nuclear war — two categories already treated as societal-scale by governments and international bodies.

That framing matters for how you read it. The statement is a coordination device: it gives researchers, policymakers, and the public a shared reference point for treating extreme AI risk as legitimate rather than speculative.

Who signed it and why that matters

The statement's weight comes from its signatories. It was signed by a large number of AI researchers and notable public figures, including many of the people building the most capable systems. The value of a broad signatory list is that it removes the "this is a fringe concern" objection — when people inside the field endorse the priority claim, the debate shifts from whether to address extreme risk to how.

If you are citing the statement, cite it as a consensus signal about priority, not as evidence about probability or as an endorsement of any particular safety approach. Signatories agree on the sentence; they do not necessarily agree on what follows from it.

How it relates to CAIS's overall mission

CAIS describes itself as a San Francisco-based research and field-building nonprofit whose mission is to reduce societal-scale risks from AI by conducting safety research, building the field of AI safety researchers, and advocating for safety standards. The extinction-risk statement sits inside that mission as the advocacy-facing piece: it states the why at the highest level of stakes.

The rest of CAIS's work is the how, and it is broader than the statement alone:

• Research — technical safety work, including the ML Safety infrastructure effort.
• Field-building — programs such as the AI and Society Fellowship, the Philosophy Fellowship, and the AI Safety, Ethics, & Society training course, which CAIS says has reached 1000+ future AI safety leaders.
• Infrastructure — a CAIS compute cluster supporting safety research.
• Education and outreach — an Introduction to AI Safety course in partnership with Coursera, covering drivers of modern AI development, risks like malicious use, competitive pressures, and loss of control, and governance strategies at different scales.

CAIS's director, Dan Hendrycks, frames the approach as multidisciplinary: preventing extreme risks from AI "requires more than just technical work," which is why CAIS works across academic disciplines, public and private entities, and the general public.

The FAQ and the objections it addresses

CAIS maintains a Frequently Asked Questions section alongside the statement. This is where the predictable pushback gets handled — questions like whether the statement is overstating the risk, whether signing implies a specific regulatory position, and how extinction risk relates to nearer-term harms. If you are evaluating the statement for a paper, a policy discussion, or a class, read the FAQ before quoting the sentence, because the FAQ is where CAIS scopes what the statement does and does not commit signatories to.

How to use this if you are deciding something

• If you need a citable consensus signal that extreme AI risk is treated as a global priority by people in the field, the statement is the right artifact.
• If you need an argument or evidence that extinction risk is high, the statement is not that — it is a priority claim, and CAIS's research pages are the place to look for technical substance.
• If you are deciding whether to engage with CAIS, note that involvement runs through several distinct channels: research, fellowships and training, the compute cluster, and public education. Which one fits depends on whether you are a researcher, a student, or working on policy.

The statement's role is to set the stakes; CAIS's research, field-building, and training programs are where the organization acts on them.