EA Forum Podcast (All audio)

This is a brief overview of the Center on Long-Term Risk (CLR)'s activities in 2025 and our plans for 2026. We are hoping to fundraise $400,000 to fulfill our target budget in 2026.

About us

CLR works on addressing the worst-case risks from the development and deployment of advanced AI systems in order to reduce s-risks. Our research primarily involves thinking about how to reduce conflict and promote cooperation in interactions involving powerful AI systems. In addition to research, we conduct a range of activities aimed at building a community of people interested in s-risk reduction, and support efforts that contribute to s-risk reduction via the CLR Fund.

2025 was a year of significant transition for CLR. Jesse Clifton stepped down as Executive Director in January, succeeded by Tristan Cook and Mia Taylor as Managing Director and Research Director. Following Mia's subsequent departure in August, Tristan continues as Managing Director with Niels Warncke leading empirical research.

During this period, we clarified the focus of our empirical and conceptual research agendas: respectively, studying the emergence of undesirable personas in LLMs, and developing interventions to get AIs to use “safe Pareto improvements” to prevent catastrophic [...]

---

Outline:

(00:31) About us

(01:48) Review of 2025

(01:52) Research

(04:50) Community building

(05:19) Donate

(05:59) Get involved

---

First published:

Source:

Linkpost URL:
https://longtermrisk.org/annual-review-fundraiser-2025/

---

Narrated by TYPE III AUDIO.

Summary

In this post, we provide an overview of responses to the cause area prioritization questions in the 2024 EA Survey. We report on the relative ranking of cause areas, how this has changed over time, and how prioritization relates to respondents’ preferred resource allocation between causes. We also consider how other responses, such as endorsement of ideas related to EA, engagement, and demographic factors, relate to prioritizations.

Overall cause prioritization

• AI risks and global health and poverty were considered the most highly prioritised causes, followed by biosecurity, and, unlike previous years, animal welfare
• Animal welfare and AI risks received higher prioritisation ratings than in the last survey year
• High-engagement respondents rated AI risks, EA movement building, and animal welfare higher
• Low-engagement respondents rated climate change and mental health higher
• High engagement was the strongest predictor of cause prioritization 

Philosophical ideas related to cause prioritization

• The agreement or disagreement with certain ideas was related to cause prioritizations in domains such as animal welfare, AI risks, global health and poverty, and climate change
• Cause areas relating to bio- and nuclear security and cause prioritization research were not related to agreement with any ideas
• Agreement has changed very little [...]

---

Outline:

(00:12) Summary

(00:43) Overall cause prioritization

(01:19) Philosophical ideas related to cause prioritization

(01:59) Allocation of resources to causes

(02:51) Cause Prioritizations

(06:20) Trends in cause prioritization over time

(06:52) en-US-AvaMultilingualNeural__ Line graph titled Trends in cause prioritization over time showing multiple causes from 2015 to 2024.

(08:48) Engagement and cause prioritization

(10:01) Factors linked with prioritization of different causes

(11:39) EA-related Ideas and Cause Prioritizations

(14:40) Time trends in EA-related ideas endorsement

(15:32) Relationship between ideas and causes

(16:54) Allocation of resources across causes

(18:17) By Engagement

(19:03) Factors linked with resource allocation

(20:25) Acknowledgments en-US-AvaMultilingualNeural__ Rethink Priorities logo

(20:38) Appendix

(20:41) Methodological notes on cause prioritization, linear regressions

(24:09) Additional partial dependence plots for cause prioritization linear regressions

(24:28) EA-related ideas agreement

(24:33) en-US-AvaMultilingualNeural__ A bar graph showing EA ideas agreement by engagement across different categories.

(24:54) Methodological notes on resource allocation linear regressions

(25:30) Associations between resource allocations and cause prioritization ratings

---

First published:

Source:

---

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

Cage-free layer campaigns are one of the most celebrated success stories in animal welfare and in EA-aligned work, improving the lives of hundreds of millions of hens and reshaping entire global supply chains. But alongside that incredibly important work, few people realize that an even larger and far more neglected confinement system exists almost entirely out of sight: broiler (meat) chickens kept in stacked wire cages, primarily in China.

• Scale: China raises ~15 billion broiler chickens annually, and ~10 billion (60–70%) spend their lives in cramped cages with even less space than battery-caged layers. This is 2x the total number of caged laying hens worldwide, with severe welfare implications.
• Neglectedness: Despite the enormous scale, this issue has received virtually no attention from the animal-welfare or EA communities. Broiler cages in China were almost entirely unaddressed, with effectively no corporate commitments on record before this year.
• Tractability: This may be one of the most overlooked high-impact opportunities in farmed-animal welfare.
  • Roughly half of major Chinese producers already offer cage-free broiler chicken.
  • Cage-free broiler meat is at price parity with caged meat.
  • Major restaurant and retail chains have already begun committing to cage-free sourcing.

These conditions create a [...]

---

Outline:

(00:13) TL;DR

(02:54) The Scale of the Problem

(04:21) Why This Looks Unusually Tractable

(04:32) 1. Cage-Free Broiler Chicken Is Widely Available

(05:19) 2. Price Parity: Switching Costs Are Essentially Zero

(06:35) 3. Early Market Momentum

(07:57) Who We Are and What Additional Funding Would Enable

---

First published:

Source:

---

Narrated by TYPE III AUDIO.

---

First published:

Source:

---

Narrated by TYPE III AUDIO.

Anthropic is untrustworthy.

This post provides arguments, asks questions, and documents some examples of Anthropic's leadership being misleading and deceptive, holding contradictory positions that consistently shift in OpenAI's direction, lobbying to kill and water down regulation so helpful that employees of all major AI companies speak out to support it, and violating the fundamental promise the company was founded on. It also shares a few previously unreported details on Anthropic leadership's promises and efforts.[1]

Anthropic has a strong internal culture that has broadly EA views and values, and the company has strong pressures to appear to follow these views and values as it wants to retain talent and the loyalty of staff, but it's very unclear what they would do when it matters most. Their staff should demand answers.

I would like to thank everyone who provided feedback on the draft; was willing to share information; and raised awareness of some of [...]

---

Outline:

(01:37) 0. What was Anthropics supposed reason for existence?

(05:04) 1. In private, Dario frequently said he won't push the frontier of AI capabilities; later, Anthropic pushed the frontier

(10:59) 2. Anthropic said it will act under the assumption we might be in a pessimistic scenario, but it doesn't seem to do this

(15:06) 3. Anthropic doesnt have strong independent value-aligned governance

(15:12) Anthropic pursued investments from the UAE and Qatar

(18:01) The Long-Term Benefit Trust might be weak

(18:35) More general issues

(19:43) 4. Anthropic had secret non-disparagement agreements

(22:27) 5. Anthropic leaderships lobbying contradicts their image

(24:34) Europe

(25:13) SB-1047

(34:33) Dario argued against any regulation except for transparency requirements

(35:08) Jack Clark publicly lied about the NY RAISE Act

(37:08) Jack Clark tried to push for federal preemption

(37:33) 6. Anthropics leadership quietly walked back the RSP commitments

(38:24) Unannounced removal of the commitment to plan for a pause in scaling

(39:20) Unannounced change in October 2024 on defining ASL-N+1 by the time ASL-N is reached

(41:01) The last-minute change in May 2025 on insider threats

(41:40) 7. Why does Anthropic really exist?

(47:37) 8. Conclusion

---

First published:

Source:

Linkpost URL:
https://anthropic.ml

---

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.

Crossposted from LessWrong.

TL;DR: Figure out what needs doing and do it, don't wait on approval from fellowships or jobs.

If you...

• Have short timelines
• Have been struggling to get into a position in AI safety
• Are able to self-motivate your efforts
• Have a sufficient financial safety net

... I would recommend changing your personal strategy entirely.

I started my full-time AI safety career transitioning process in March 2025. For the first 7 months or so, I heavily prioritized applying for jobs and fellowships. But like for many others trying to "break into the field" and get their "foot in the door", this became quite discouraging.

I'm not gonna get into the numbers here, but if you've been applying and getting rejected multiple times during the past year or so, you've probably noticed the number of applicants increasing at a preposterous rate. What this means in practice is that the "entry-level" positions are practically impossible for "entry-level" people to enter.

If you're like me and have short timelines, applying, getting better at applying, and applying again, becomes meaningless very fast. You're optimizing for signaling competence rather than actually being competent. Because if you a) have short timelines [...]

---

First published:

Source:

---

Narrated by TYPE III AUDIO.

The (new) Center for Wild Animal Welfare, the Wild Animal Initiative, the EA Animal Welfare Fund and Animal Charity Evaluators have all been in the Donation Election's top 3 candidates this week. How should we compare between them?

I don't have a settled take on this question, and I won't be giving an argument (at least initially). I'm looking forward to reading your takes.

Some considerations:

• There are many more wild animals than farmed animals, especially if we count invertebrates.
• It's likely less than $10M is spent on wild animal welfare annually, whereas around $200M is spent on farm animal welfare. (These are both extremely low compared to the size of the problem).
• It's often argued that we shouldn't get involved in wild animal welfare because we don't know what kind of complex ecological effects we might cause. Unfortunately, this is true of most things, including farm animal welfare. 

I've framed the debate around a 'marginal $5K' because, at time of writing, that's roughly what is at stake if an organisation is in/ not in the top three Donation Election candidates by the end of the week.

---

First published:

Source:

---

Narrated by TYPE III AUDIO.

Acknowledgments: Tom Marty, Vanessa Sarre, Clare Diane Harris and Jack Koch contributed to the design of the survey, Jack Koch also carried out outreach to patient support groups, and Magdalena Kolczyńska, Jordan Clist and Michael Smith contributed to the analysis of the raw data. We sincerely thank every individual who took the time to respond to the survey and reflect on their own suffering.

• Existing wellbeing metrics, including the DALY and measures of improved life satisfaction such as the WELLBY, do not directly measure suffering – a key determinant of wellbeing – and suffering metrics can address this limitation.
• A pilot survey we conducted with 485 respondents generated retrospective estimates of suffering duration and intensity for a range of conditions and life experiences over one-year periods, as well as momentary wellbeing assessments on 0-10 scales.
• While retrospective estimates almost certainly overstated many durations, they also captured extreme suffering that may have been underrepresented in real-time evaluations.
• Even common conditions, such as tinnitus and loneliness, can cause suffering that is evaluated as extreme, and the results suggest that several other cause areas may deserve greater attention due to the intensity of the suffering.
• As a next step, a [...]

---

Outline:

(00:46) Summary

(02:09) The need for suffering metrics

(05:10) Pilot suffering survey

(08:54) Findings

(08:57) Psychological suffering most commonly experienced

(09:35) Conditions most & least frequently selected as major source of suffering among those ever experienced

(11:56) Descriptions

(23:01) Duration estimates at different suffering intensities

(24:52) Insights from a few selected conditions

(24:57) Cluster headache - a deeper analysis

(30:17) Tinnitus

(31:20) Loneliness

(31:49) Kidney stones

(32:19) Akathisia

(32:56) Momentary wellbeing correlations

(35:54) Conclusions and improvements for next, larger-scale iteration

---

First published:

Source:

---

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.

MIRI is running its first fundraiser in six years, targeting $6M. The first $1.6M raised will be matched 1:1 via an SFF grant. Fundraiser ends at midnight on Dec 31, 2025. Support our efforts to improve the conversation about superintelligence and help the world chart a viable path forward.

MIRI is a nonprofit with a goal of helping humanity make smart and sober decisions on the topic of smarter-than-human AI.

Our main focus from 2000 to ~2022 was on technical research to try to make it possible to build such AIs without catastrophic outcomes. More recently, we’ve pivoted to raising an alarm about how the race to superintelligent AI has put humanity on course for disaster.

In 2025, those efforts focused around Nate Soares and Eliezer Yudkowsky's book (now a New York Times bestseller) If Anyone Builds It, Everyone Dies, with many public appearances by the authors; many conversations with policymakers; the release of an expansive online supplement to the book; and various technical governance publications, including a recent report with a draft of an international agreement of the kind that could actually address the danger of superintelligence.

Millions have now viewed interviews and appearances with Eliezer and/or Nate [...]

---

Outline:

(02:18) The Big Picture

(03:39) Activities

(03:42) Communications

(07:54) Governance

(12:31) Fundraising

---

First published:

Source:

---

Narrated by TYPE III AUDIO.

In the past 10 years, Coefficient Giving (formerly Open Philanthropy) has funded dozens of projects doing important work related to AI safety / navigating transformative AI. And yet, perhaps most activities that would improve expected outcomes from transformative AI have no significant project pushing them forward, let alone multiple.

This is mainly because we and other funders in the space don't receive promising applications for most desirable activities, and so massive gaps in the ecosystem remain, for example:[1]

The original text contained 2 footnotes which were omitted from this narration.

---

First published:

Source:

---

Narrated by TYPE III AUDIO.