Project summary

AI governance work in consumer states tends to run ahead of the technical grounding behind it. People doing policy and legislative work in places like the Philippines are often building on alignment research findings without having the evaluative training to know which of those findings still hold under scrutiny. The introductory technical programs I went through (e.g. BlueDot, Condor Initiative, GCP) were designed to map the major open problems in alignment. Examining how the foundational arguments behind them have aged under a decade of scaling evidence is a later-stage kind of training, and I have come to notice that the gap it leaves is not obvious at first. I have observed this gap most clearly in retrospect, when policy work already built and underway turns out to rest on premises that researchers have since revised or abandoned.

I am a core member of AI Safety Diliman, previously supported by the Berkeley Existential Risk Initiative earlier this year. That work grew out of advising Philippine Congress on early AI legislation in 2024 and now continues through our AI Governance Career Accelerator, which trains early-career practitioners across two pathways, non-frontier governance for consumer states like the Philippines and frontier engagement with global AI safety institutions. Several fellows from the program are now engaged in drafting Philippine AI legislation through the House Committee on ICT.

This trajectory is what brings me to Europe this July. I will attend the inaugural Global Dialogue on AI Governance at the UN and the AI for Good Global Summit in Geneva from July 3 to 11, with the Heinrich Böll Foundation covering my international travel, accommodation, and per diem for that portion of the learning experience. I was separately and very recently accepted to the Human-Aligned AI Summer School (HAAISS) in Prague immediately after. HAAISS has generously covered the school fee and accommodation through their financial aid, but travel falls outside what they're able to support for me. This request is for the transportation between Geneva, Prague, and my return home, that I cannot cover on my own.

What are this project's goals? How will you achieve them?

I am hoping to attend HAAISS in full, from July 13 to 16, without arriving late or having to leave early because the transportation gap could not be closed. Two tracks matter most deeply to me. The arguments-for-AI-risk-revisited track examines how mesa-optimization, scheming, and discontinuity risk have fared against a decade of scaling evidence, and I look forward to this, because I currently know what these arguments claim, but not how researchers who have lived through that evidence have updated their thinking, and that distinction shapes whether the technical governance work I care about is built to last or already standing on shaky ground. The complex systems track would let me sit with ACS Research's account of gradual disempowerment alongside the people who developed it, especially the question of how disempowerment moves through multi-agent ecosystems and which institutions end up resisting it versus making it worse. The original paper leaves this open, and it is a question I have carried since first reading it in 2025, back when I was drafting an undergraduate thesis proposal around it before my college decided AI was not yet a priority research area. Being in a room with the people who built the framework, instead of working through it alone against the same papers I have already read closely, is the difference I am hoping this summer school makes, as I begin preparing graduate school applications later this year.

Getting there is mostly a matter of logistics now, as I have already been accepted to the summer school, and I am already on my way to Europe for the first time for technology conferences and gatherings in Geneva right before the summer school begins. The only thing left standing between me and actually being in the summer school fully is the cost of the trains and the flight change, and I would be deeply grateful for help covering it.

How will this funding be used?

My return flight home was already booked by the Foundation, departing from Barcelona. After several exchanges with the airline, I found it more economical to keep the original Barcelona departure instead of paying the fee to reroute the entire itinerary through Prague. I have already covered the Geneva to Prague bus and a connecting flight from Prague to Barcelona using my own funds, and this request includes reimbursement for those costs alongside the remaining unpaid balance, the airline's date change fee to shift my Barcelona departure from July 16 to July 17.

• Geneva to Prague, direct bus: ~$104

• Local transport in Prague during the summer school and Prague to airport transfer via public transportation: ~$15

• Budget flight from Prague to Barcelona, including checked luggage: ~$174

• Fare difference to shift Barcelona departure from July 16 to July 17, confirmed with the airline: ~$268

Total cost: ~$561

At the $500 minimum, most of what I've already spent out of pocket would be reimbursed, but the fare difference fee would remain a gap I'd have to absorb on a student budget that has very little room for it. At full funding of $561, that fee is covered too, and I wouldn't be carrying any of this cost on my own.

Who is on your team? What's your track record on similar projects?

As this is an individual request, I am the sole beneficiary of the funding.

My track record includes Condor Initiative's AI Safety Camp in Southeast Asia, an immersive governance and technical alignment program with direct engagement from alignment researchers and field-builders, and Global Challenges Project's Workshop on Existential Risk in 2023. I co-founded Effective Altruism UP Diliman and continue helping sustain AI Safety Diliman as an ongoing organization. I am also a graduating undergraduate serving as senior research associate to the national university's Vice President for Digital Transformation.

What are the most likely causes and outcomes if this project fails?

The one risk specific to this routing is that the Prague to Barcelona leg and the Barcelona to Dubai leg are separate bookings on different airlines, with no protection between them if the first flight is delayed or cancelled. I am building in same-day buffer time to reduce this risk, but it is the one point of fragility in this plan. If the Prague to Barcelona flight were disrupted badly enough to threaten the airlines' connection, I would need to book a same-day alternative at my own cost, which would be difficult to absorb given the relatively low fares on this specific route.

Beyond that operational risk, the line items above are otherwise confirmed directly with the relevant providers.

How much money have you raised in the last 12 months, and from where?

I personally have not raised programmatic funding before this request. AI Safety Diliman as an organization received support from the Berkeley Existential Risk Initiative earlier this year, but that funding was raised by our president, and not by me individually for a personal request.