EA Forum Podcast (All audio)

This is a link post.

These days, I often run across whippersnappers excited to do something for AI safety — but aren’t quite sure what. One of the fun things about the Future Fund era were the big lists of project ideas; as we enter a new era of crazy money sloshing around, it might be time to bring back the lists!

Note that these ideas range from “very confident this is good” to “completely harebrained”; I’m not telling you which are which.

If you’re excited for ideas like these, consider joining Surplus, our upcoming software incubator: https://manifund.org/surplus

Jobs, jobs, jobs

Already, the top problem for most AI safety orgs is hiring good people. Vast torrents of funding will only exacerbate the imbalance between available money and people to hire. So now is a great time to figure out how to discover new talent & match people to jobs.

1. Triplebyte for AI safety jobs

Triplebyte would interview a tech candidate once, then forward the results to a bunch of different companies. This would reduce the O(MN) problem in hiring between M orgs and N people to O(M+N), saving applicants and interviewers time. Most obviously, you could just do this [...]

---

Outline:

(00:44) Jobs, jobs, jobs

(01:06) 1. Triplebyte for AI safety jobs

(01:55) 2. Database of every single AI safety person

(02:48) 3. Intro AI safety megaconference

(03:50) 4. Unionize the lab employees

(05:22) 5. Safety scorecards to inform job applicants

(06:11) 6. Help people move to US (and maybe Bay Area specifically)

(06:47) Cargo-culting from academia

(07:13) 7. AI Safety Nobel Prize

(08:27) 8. University of AI Safety

(09:09) 9. Good AI safety research journal

(09:33) 10. Legit, academic AI safety conference

(10:07) Products I would consume

(10:22) 11. Turn the AI 2027 TTX into a mass market product

(11:29) 12. "Lighthaven/Constellation/Mox in DC"

(11:37) 13. Common application for AI safety funding

(11:57) 14. Givewell for AI safety

(12:21) 15. In-depth profiles of AI safety leaders

(13:16) 16. "what-is-agi.com"

(14:16) Questions to generate project ideas

(16:11) Other notes and advice

---

First published:

June 3rd, 2026

Source:

https://forum.effectivealtruism.org/posts/EvfQ8Dy5ewXGTnEyv/sixteen-schemes-for-ai-safety

Linkpost URL:
https://manifund.substack.com/p/sixteen-schemes-for-ai-safety

---

Narrated by TYPE III AUDIO.

TLDR: The Open Wing Alliance (OWA) just launched a global cage-free campaign against the largest Korean food multinational, CJ Group. Thanks to a once-in-a-generation alignment of factors, any action you take to support this campaign is likely to have an unusually high expected value, easily as high as the recent Ahold Delhaize campaign.

I'm the Data and Insights Lead at The Humane League. Here's the case.

Why Korea, why now

South Korea is making a generational infrastructure decision. A September 2027 ban on the smallest (500cm2) battery cages means the Korean egg industry is in the process of investing millions into new housing systems. That means either slightly larger cages or aviary systems. Those decisions are being made as I write, and the window will close in the next approximately equals 6-9 months (as new farms take several months).

The conditions for a cage-free transition are an order of magnitude more favourable than anything I've seen in an Asian market. Cage-free retail egg sales in Korea have tripled since 2022 despite a 58% price premium and no commitments from the top retailers - a level of consumer willingness to pay unmatched anywhere in Asia and possibly the world. Some [...]

---

Outline:

(00:39) Why Korea, why now

(02:25) Why CJ

(03:21) Why the expected value is enormous

---

First published:

June 3rd, 2026

Source:

https://forum.effectivealtruism.org/posts/tk4phm3ASHuhtXgtM/an-ahold-sized-opportunity-to-accelerate-cage-free-in-asia

---

Narrated by TYPE III AUDIO.

I used AI in this post. >30% is AI-generated text (everything that appears in quote blocks).

Google's Debug Project has been releasing mosquitos infected with bacteria that stop them from breeding in the wild, as a way of reducing mosquito populations.

This is a promising strategy for fighting mosquito-borne disease; Google's particular approach doesn't target malaria, but it has been shown to reduce rates of dengue fever (and similar approaches with gene drives can address malaria; the linked organization has been supported heavily by Coefficient Giving).

The Debug Project wants to try their approach with a new species of mosquito, and they've requested permission from the EPA. If they get it, they plan to release up to 64 million mosquitos across Florida and California.

Public comments are open through June 5. Predictably, they've been mostly negative so far, because the optics of "releasing millions of infected mosquitos" are poor. (The number might be ~100%, because I read 15 comments and all 15 were negative.)

However, past releases in places like Colombia and Fresno, California show that this project is likely to be highly effective. And I'd guess that the more projects like this happen, the more it will [...]

---

First published:

May 31st, 2026

Source:

https://forum.effectivealtruism.org/posts/MSEJXCXuiHaJDcekT/a-chance-to-support-large-scale-mosquito-eradication-in-the

---

Narrated by TYPE III AUDIO.

Air Filters for Infection

Air filters are often proposed as a simple, scalable way to reduce the spread of respiratory infections. The underlying logic is simple - respiratory pathogens can be transmitted through the air in aerosol form, and air filters are capable of removing these particles before they are inhaled and go on to cause infection. This has led to sustained interest in air filtration as a non-pharmaceutical intervention to prevent the spread of respiratory disease.

I’ve spent a significant portion of my time at 1Day Sooner researching interventions for indoor air quality. Throughout that time, I’ve become increasingly bearish about the utility of air filters for preventing respiratory infection. I argue that, because respiratory infections are driven by short-range exposure to highly concentrated emissions and because empirical findings also fail to find a link between filters and infectious disease, air filters are unlikely to work well in a wide variety of settings and situations.

Summary

• The theoretical argument that air filters significantly prevent the spread of infectious disease is weak. Much infectious disease transmission occurs over short times and distances, which air filters fail to disrupt.
• Evidence for the real-world effectiveness of filters is confined to [...]

---

Outline:

(00:11) Air Filters for Infection

(01:11) Summary

(01:48) Real-world studies

(05:32) CAFACS (Clean Air For Aged Care Study)

(09:33) AFRI-c

(10:43) Theoretical basis for filtration

(13:36) Conclusion

---

First published:

May 28th, 2026

Source:

https://forum.effectivealtruism.org/posts/phtZxxPyP9tizQQaF/why-i-m-skeptical-of-air-filters-for-infection-prevention

---

Narrated by TYPE III AUDIO.

---

Images from the article:

Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.

Crosspost.

Billions of dollars in philanthropic funding is coming down the pipeline, and we’re not ready for it.

Nan Ransohoff recently released a hugely important piece titled The third wave of American philanthropy. In the relatively near future, three philanthropic behemoths will have a huge influx of liquid cash: the OpenAI foundation, Anthropic's founders, and Anthropic employees.

A fairly conservative estimate puts the amount of charitable spending this would open up at around 50 billion dollars per year. This is about 1/2000th the annual GDP of Earth. US charitable giving is about 600 billion dollars per year—so this will increase total US charitable giving by about 8%. Crucially, many of these donors are interested in sponsoring high-impact charitable projects. Ransohoff notes:

50 billion dollars/year could fully fund the annual budgets of the following organizations:

• 6 Gates Foundations (~9 billion dollars/yr), or
• 67 Coefficient Givings (~1 billion dollars/yr), or
• 100 GiveWells (~500 million dollars/yr), or
• 333 Arc Institutes (~150 million dollars/yr), or
• 5000 Institutes for Progress (~10 million dollars/yr)

Obviously, there are many effective organizations beyond those listed here. But the takeaway is that we are orders of magnitude off from having the great organizations needed to absorb [...]

---

First published:

June 2nd, 2026

Source:

https://forum.effectivealtruism.org/posts/PjnRtq8intQdeeDfA/philanthropy-needs-ambitious-projects-immediately

---

Narrated by TYPE III AUDIO.

This post was originally published on the GiveWell blog. You can view the original version here.

This year, our research team is focused on two primary goals. The first is to scale our capabilities so we’re able to move much more donor funding to highly cost-effective programs in the next few years. The second is to grant at least $500 million to the best opportunities we can find this year to save and improve lives.

Over the past several years, GiveWell has doubled the size of our research team to deepen and broaden our search for highly impactful programs. Our 60 researchers are now distributed among 11 subteams that cover a number of global health and development cause areas, as well as core research needs.

Below you’ll find a summary of the key approaches each subteam is using this year to find new opportunities to help people in need as much as we can.

Malaria

People: Marinella Capriati, Zoe Hartman

Malaria, which is caused by a parasite transmitted when people are bitten by infected mosquitoes, is a leading cause of death globally, especially for young children in Africa, who make up around 70% of the approximately 600,000 malaria deaths [...]

---

Outline:

(01:04) Malaria

(01:48) Antimalarial Medicines

(03:55) Malaria Vector Control

(05:44) Malaria Cross-Cutting Research

(07:31) Nutrition

(10:12) Vaccination

(12:25) Water

(15:23) Livelihoods

(17:37) New Areas

(19:57) Cross-Cutting Research

(22:34) Research Support and Operations

(22:55) Research Operations

(23:52) Commons

---

First published:

June 1st, 2026

Source:

https://forum.effectivealtruism.org/posts/LotTxJF3PiHw6iLDq/how-we-re-searching-for-the-best-ways-to-help-in-2026

---

Narrated by TYPE III AUDIO.

TL;DR: You should run a virtual summer Intro Program if you're in college. It's an easy way to boost an existing group or (re)start one. Most of the resources you need are already available, and I am here to help with planning or advice, even if you've never done any community building before!

Princeton EA already received over 130 sign-ups for their virtual summer 2026 Intro Fellowship – with 10 facilitators running 14 cohorts! Harvard, Brown, MIT, and other schools are testing the same model, and you should too!

First, some background: One year ago, in the summer of 2025, Yale EA advertised a Summer Introductory Fellowship. The results were highly promising. They started off the year noticeably stronger than many other EA groups, with regular attendance at weekly meetings, and a few summer participants joining the board early on.

Now, more schools are testing summer fellowships. I am extremely excited about this opportunity and think all university students should be too. Your group can get a huge head start by being organized enough to do it. Incoming freshmen are eager to make friends and don't have any existing ties to particular clubs on campus, which means you have [...]

---

First published:

May 29th, 2026

Source:

https://forum.effectivealtruism.org/posts/kjh88tgtqhYWuFmau/college-students-you-should-run-a-virtual-summer-ea-intro

---

Narrated by TYPE III AUDIO.

[Cross posted from my substack]

In their EA Forum post last year, CEA described their ‘principles-first approach to stewardship of the EA community’.

I'm a big fan of principles-first stewardship in principle. I think EA needs a steward, and I think that stewardship should be organised around EA's core principles.

But I think CEA's particular growth-centric approach to principles-first stewardship is stewarding EA in the wrong direction.

I think that: The key question for principles-first stewardship should be "Is EA a place that embodies and nurtures EA principles?" I think there are serious reasons to worry that it isn't such a place - that EA has become more ideological and less truth-seeking over time, and that growth focused approaches to community building like CEAs are a big part of the reason why.

A summary of my main points:

• It seems to me that EA is dying. I’m less concerned here about growth metrics, and more concerned about the health of EA as a community and a moral/ intellectual project. It seems to me that EA is losing its question-nature, and also has become something that people are less and less willing to stand behind or participate [...]

---

Outline:

(04:16) It seems to me that EA is dying

(05:05) EA as a question

(06:26) EA as a community

(07:30) Various posts which inform my conception of EA death

(08:42) Growth is not "Community Building 101"

(10:20) The growth funnel model is in tension with open truth-seeking

(10:25) Targeting high impact careers and donations

(13:04) Selection effects

(15:20) Growth is only good if EA is functioning well

(17:23) EA community building doesn't serve the people who embody EA most deeply

(20:01) FTX was a trust problem, not just a brand problem

(22:14) CEA's brand strategy is in tension with open truth-seeking

(24:33) What principles-first stewardship could look like

---

First published:

May 28th, 2026

Source:

https://forum.effectivealtruism.org/posts/xsffhcHoexJgH4h4X/my-disagreements-with-cea-s-approach-to-stewarding-ea

---

Narrated by TYPE III AUDIO.

I have linked below my recent version of my research compilation on Profit for Good businesses and the Charitable Ownership Advantage thesis. I have spent several hundred, if not over a thousand hours, compiling the evidence supporting the thesis that, given our modern economy in which ownership is typically practically separate from business management and governance, Profit for Good businesses should outperform conventionally owned businesses due to non-zero preferences by stakeholders such as consumers and employees, who affect business performance. I have pasted below the Introduction and Executive Summary to this compilation, but even if you would only like to read that (which is probably enough for most people), I would encourage you to read it through the below hyperlink, given that there were some issues with the images (graphs, tables, etc.) being copied over. I note that although I reviewed the sources and iterated extensively regarding drafting the compilation, AI was used extensively in virtually all parts of the process, including drafting.

Link to the entire Research Compilation 

Profit for Good

and the Charitable

Ownership Advantage

Brad West

Project COA

May 2026

Table of Contents

Executive Summary............................................................................................................................ 1

Section 1: Stakeholder Preferences...................................................................................... 34

1.1 Consumer Preferences...................................................................................................... [...]

---

Outline:

(01:08) Link to the entire Research Compilation

(01:24) List of Exhibits

(01:28) Executive Summary

(01:58) Section 1: Stakeholder Preferences

(02:19) Section 4: Charitable Ownership Advantage

(02:39) Section 6: Preference and Charitable Commitment Concentration

(03:09) Section 7: Trust and Awareness Infrastructure

(03:22) Section 8: Stress Tests

(03:34) Section 9A: The First Proof Portfolio

(04:37) Section 9B: Sector Strategy and Financial Projections

(05:01) How to Read This Compilation

(09:35) Profit for Good and the Charitable Ownership Advantage

(17:39) The case in one page

(20:50) Key terms

(25:36) 1. Headline findings and action agenda

(37:22) From first acquisitions to category scaling

(38:16) The strategic asymmetry

(41:21) 2. What we know, what we are testing, what would change our mind

(41:27) What is well-supported

(44:37) What remains empirical

(46:17) What would update the thesis

(51:53) What would falsify or redirect the thesis

(52:47) 3. Recommendations: build the proof object and choice infrastructure

(53:31) Two projects

(56:32) Capital architecture

(58:10) Best-fit roles by funder type

(59:22) 4. The strategic moment

(01:04:01) Section-by-Section Synthesis

(01:04:30) §1.1 -- Consumer Preferences

(01:06:13) §1.2 -- Employee Preferences

(01:07:58) §1.3 -- Supplier and Partner Preferences

(01:09:57) §1.4 -- Media and Marketing Advantages

(01:11:56) §1.5 -- Capital Provider Preferences

(01:14:06) §1.6 -- Other Stakeholder Preferences (B2B, Government, Community)

(01:16:22) §2.0 -- Direct PFG Evidence

(01:19:01) §3.0 -- Global Expansion Opportunity

(01:21:21) §4.0 -- Charitable Ownership Advantage: Parity, Redeployability, and Deployment as Strategic Choice

(01:24:02) §5.0 -- The Awareness-Trust Activation Dynamic

(01:26:34) §6.0 -- Preference and Charitable Commitment Concentration

(01:30:08) §7.0 -- Trust and Awareness Infrastructure

(01:33:32) §8.0 -- Stress Tests

(01:36:38) §9A -- Capital Deployment: The First Proof Portfolio

(01:39:40) §9B -- Sector Strategy and Financial Projections

(01:44:36) §10 -- Synthesis and Recommendations

---

First published:

May 28th, 2026

Source:

https://forum.effectivealtruism.org/posts/CJJjhvsRafGFn6S6v/profit-for-good-390-page-research-compilation

---

Narrated by TYPE III AUDIO.

---

Images from the article:

Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.

TL;DR: Anthropic restricted access to Claude Mythos Preview, citing a major leap in vulnerability discovery and exploitation capability. I review the 3 most common arguments from skeptics: (1) AISLE Security's paper showing cheaper models can identify the same bugs as Mythos, (2) benchmark comparisons showing GPT-5.5 performs comparably, and (3) Mythos finding only one low-severity bug in the cURL project.

• On most cyber capabilities: the skeptics are right. Mythos isn’t dramatically ahead of GPT-5.5, and GPT-5.5 is more cost-efficient for most use cases.
• On vulnerability discovery and exploitation capabilities specifically: the skeptics are wrong, or at least overreaching. AISLE Security's results don’t replicate when models are tested under similar conditions (Ex: Semgrep's experiment), and benchmarks designed to actually measure vulnerability discovery and exploitation skills (XBOW AI, ExploitBench) show Mythos substantially ahead.
• The cURL result is real but not decisive: Firefox and Palo Alto Networks have reported the opposite pattern. More data is needed before we can draw a definitive conclusion.

(For context, my background is in penetration testing and bug bounty hunting, mostly specializing in web security, secure code review, and cloud security.)

AI has been measurably accelerating vulnerability research. The volume of reported software vulnerabilities continues to [...]

---

Outline:

(04:04) #1 AISLE Security's Paper

(10:10) #2 Mythos vs. GPT 5.5 Benchmark Performance

(17:07) #3 Only 1 Vulnerability Found in cURL

(20:12) Conclusion

---

First published:

May 23rd, 2026

Source:

https://forum.effectivealtruism.org/posts/8yztpbjuPkyXsmA6n/are-mythos-cyber-capabilities-overstated-yes-and-no

---

Narrated by TYPE III AUDIO.

---

Images from the article:

Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.