PIBBSS Fellowship 2026–2027: Historical Reference for the Closed AI Safety Research Program in Cape Town
Historical reference for the closed 2026–2027 Winter PIBBSS Fellowship, an approximately three-month interdisciplinary AI safety research program in Cape Town with a $3,000 monthly stipend, accommodation, meals, and a return flight.
PIBBSS Fellowship 2026–2027: Historical Reference for the Closed AI Safety Research Program in Cape Town
The PIBBSS Fellowship was an interdisciplinary research program run by Principles of Intelligence, the organization that grew out of the Principles of Intelligent Behavior in Biological and Social Systems initiative. Its purpose was to help researchers from fields that study intelligence, complex systems, institutions, and living systems move toward AI safety research with support from an experienced mentor. The official fellowship page now states that applications for the 2026–2027 Winter Fellowship are closed. This page is therefore a historical reference to that cycle, not an active application listing.
The official page does not announce a new fellowship cycle. It points readers toward an expression of interest for future opportunities, so prospective applicants should not treat the details below as an invitation to submit a late application. The closed cycle’s deadline remains recorded as July 20, 2026, because that is the date published by Principles of Intelligence and because an archive entry should preserve the actual date rather than replace it with an invented future deadline.
Key details from the closed cycle
| Item | Officially published detail |
|---|---|
| Program | PIBBSS Fellowship, 2026–2027 Winter Fellowship |
| Organizer | Principles of Intelligence |
| Status | Applications closed; no next cycle announced on the official page |
| Format | Full-time research with a dedicated mentor, primarily in person |
| Location | Cape Town, South Africa; remote participation only for exceptional candidates |
| Duration | Approximately three months, November 2026–February 2027 |
| Stipend | $3,000 per month |
| Accommodation | A private bedroom for each attendee, potentially in shared accommodation |
| Meals | Workday meals at the office, including lunch, dinner, and snacks |
| Travel | One return flight to Cape Town |
| Visa support | Short-term visa support letters available |
| Cohort | Approximately 20 fellows |
| Application deadline | July 20, 2026 |
The page’s status is the central fact for anyone finding this listing through search. A past date does not mean the program itself was unreliable; it means this particular recruitment window has finished. Readers interested in a later round should look for a new announcement from Principles of Intelligence or use the organization’s future-opportunities expression of interest. Until the organization publishes a new cycle, there is no current deadline to report.
What the fellowship offered
The 2026–2027 program was designed as a roughly three-month, full-time research placement. Fellows were expected to work on a project at the intersection of their own expertise and AI safety while receiving dedicated mentorship. The official description emphasizes matching researchers with mentors suited to their experience and goals. Applicants did not need to arrive with a fully formed project: the organization said that project proposals could be developed during the interview and mentor-matching process.
The program structure included a remote pre-program reading group on AI risks, a multi-day opening retreat, a full-time research period, shared office space with talks and community events, a multi-day closing retreat, and a final symposium presentation in Spring 2027. These elements made the fellowship more than a stipend attached to independent work. The intended experience combined orientation, concentrated research time, regular contact with the wider AI safety community, and a public account of the work produced during the placement.
The financial package was substantial. The official page listed a $3,000 monthly stipend for a full-time commitment. It also listed a private bedroom for each attendee, with the possibility that the bedroom would be in a shared apartment; workday meals at the office; and one return flight for Cape Town, South Africa. Short-term visa support letters were available. Those benefits were relevant to international researchers deciding whether an in-person placement was feasible, but they should be understood as the package described for the closed cycle, not as guaranteed terms for a future round.
The official page also reported a cohort of approximately 20 fellows and described alumni working in AI safety labs, government, academia, and independent research. It highlighted 73 alumni placed in government, AI safety labs, and universities. Those figures describe the organization’s published account of the program and its alumni; they are useful context, but they do not turn this historical page into a current funding opportunity.
Who the closed cycle was intended for
Principles of Intelligence described the fellowship as suitable for researchers motivated to contribute to AI safety and experienced in fields that study complex and intelligent systems. The listed areas included mathematics; neuroscience and cognitive science; dynamical systems theory; physics; philosophy, especially philosophy of science, mind, or ethics; political and economic theory; ecology and evolutionary biology; linguistics; media studies; and the humanities. The organization also said that the list was not exhaustive.
The selection standard was broader than a particular academic credential. The official FAQs state that a PhD was not required, published papers were not required, and applicants were expected to demonstrate something like PhD-level research ability or equivalent independent work experience. Applicants did not need previous AI safety expertise, and coding was not mandatory. This made the program relevant to researchers whose strongest evidence was an independent project, thesis, draft paper, policy analysis, model, or other serious body of work rather than a conventional academic CV.
The program accepted applicants from all countries. It was primarily in person in Cape Town, with remote participation reserved for truly exceptional cases; the FAQ characterized that as roughly about one person per cohort. It was also full time, and the organization said part-time participation was not possible. These constraints were important eligibility and planning details. Someone considering a future round would need to confirm whether the location, attendance policy, benefits, and time commitment remain the same, because the official page has only confirmed them for the closed cycle.
The program also offered a Gradual Disempowerment track for up to four fellows, made possible by ACS Research. The track concerned what happens as the systems supporting human civilization gradually stop depending on human participation, and it invited work on post-AGI economics, culture, and governance. The official page particularly welcomed economists, theorists with knowledge of cultural evolution and theoretical evolutionary biology, and polymaths comfortable working across disciplines. This track was a feature of the 2026–2027 cycle and should not be assumed to recur.
How applications worked for the closed cycle
The published application process had four stages. Stage 1 was a written application containing a CV or résumé, a 600–800 word personal statement about research background and motivation, and optional but recommended past work samples. The examples of suitable samples included papers, unpublished drafts, blog posts, newspaper articles, and similar writing. GitHub repositories or code could demonstrate coding ability, but the FAQ cautioned that code alone would not demonstrate research ability.
Stages 2–4 consisted of multiple interviews and project development. The official description says that the interviews were used to discuss research interests, develop project proposals, and match applicants with mentors. Applicants did not need a clear project idea before applying. After interviews, staff identified possible mentors based on each person’s interests and goals; fellows typically had two or three mentor interviews, after which mentors and fellows submitted preference lists and the match was completed from those rankings.
The official FAQ also described work tasks used to assess research-relevant skills. Examples included writing a research proposal, analyzing papers, or critiquing scientific writing. No special preparation was required because the tasks were intended to test abilities applicants already possessed. The page said that work tasks took place in late July and generally had to be completed within that week, with case-by-case accommodations possible. Because applications are closed, these details are useful only for understanding the past selection process and for preparing if Principles of Intelligence announces a materially similar round.
What a future applicant should verify
Do not copy the closed cycle’s dates, stipend, track, or application instructions into a new application without checking the official page first. A future announcement could alter the location, schedule, mentor process, funding package, or required materials. The current official page does not provide a new deadline, so there is no responsible date to place in this listing beyond the historical deadline of July 20, 2026.
The most useful preparation would be to keep a concise research CV current, choose one or two work samples that show independent thinking, and be ready to explain how a method from your discipline could inform an AI safety question. A future applicant should also be able to describe the practical implications of a full-time, primarily in-person placement in Cape Town. Those are preparation suggestions derived from the closed cycle’s requirements, not a claim that a future application form will ask exactly the same questions.
Frequently asked questions
Can I still apply to the 2026–2027 fellowship? No. Principles of Intelligence states on the official page that applications for the 2026–2027 Winter Fellowship are closed.
Is there a new deadline? No new fellowship deadline is announced on the official page. The date recorded here, July 20, 2026, is the closed cycle’s published deadline.
Is this page an active opportunity? No. The historicalReference flag identifies it as an archive entry for a completed recruitment cycle.
Was a PhD required? No. The official FAQ said that a PhD and published papers were not required, while substantial independent research ability was expected.
Was prior AI safety or coding experience required? No. The official FAQ said that prior AI safety expertise was not required and coding was not mandatory.
Could fellows participate remotely? Only in exceptional cases. The program was primarily in person and full time in Cape Town.
Where should I look for future opportunities? Start with the official PIBBSS Fellowship page and its expression-of-interest option for future opportunities. Confirm every new date and benefit there before relying on a later announcement.
Official source
The authoritative source for this historical record is the Principles of Intelligence PIBBSS Fellowship page: https://princint.ai/programs/fellowship/. It confirms the closed status, the July 20, 2026 deadline, the program structure, the stipend and benefits, the eligibility guidance, and the application stages described above. Since no next cycle is announced there, this page deliberately keeps the real closed-cycle date in its metadata and does not present a speculative replacement deadline.