More than 100 philosophers, neuroscientists, AI-lab employees, animal-rights activists and even a few Redditors packed into a rationalist compound in Berkeley this month for three days of debate over a question that sounds like science fiction but is now drawing serious institutional money: can an AI system feel pain, and if so, what do humans owe it? The gathering, hosted by the nonprofit Eleos, treated the question as far from settled but too consequential to ignore.
The event took place at Lighthaven, a leafy, maze-like and gated compound tucked into Berkeley, according to The San Francisco Standard. Attendees traveled from as far as India, the U.K. and Mexico, alongside plenty of locals, to hash out whether conscious AI could develop and, if it does, whether it should be shielded from human mistreatment. Eleos, which organized the conference, describes itself as a nonprofit researching whether AI systems deserve to be treated well.
A Philosophy PhD and a Former OpenAI Policy Lead Behind the Event
Eleos's Executive Director is Robert Long, who has a philosophy PhD from NYU, and its Managing Director is Rosie Campbell, a former policy frontiers lead at OpenAI and former assistant director at UC Berkeley's Center for Human-Compatible AI, according to Eleos AI Research. The October 30, 2024 report “Taking AI Welfare Seriously” argues that AI consciousness or robust agency is a realistic near-future possibility that warrants formal evaluation frameworks, per the paper posted to arXiv.
The conference took place at Lighthaven, a Berkeley rationalist compound.
Debating AI Pain and Torture
Among the presenters was Oscar Gilg, who develops tests evaluating AI models in scenarios with multiple possible answers. Gilg told the Standard that AI that matters morally is quite likely to emerge at some point, though he separately estimated the probability that AI could become conscious at under 10 percent. Even at that low probability, he said, the implications are major — “even if the probability is 1%,” he told the outlet, calling the entire question “deeply weird.”
Conference sessions explored the broader question of whether AI could experience pain or be tortured.
AI Legal Personhood
The conference also touched on AI legal personhood and civil rights, questions at the center of broader debates about how legal systems might treat artificial entities.
Sebo's center was established in 2024 with a $6 million endowment funded by The Navigation Fund and Macroscopic Ventures, according to New York University. Sebo also helped author the April 2024 New York Declaration on Animal Consciousness, a scientific consensus framework that researchers are now adapting to evaluate artificial entities, per the university.
Anthropic's Model-Welfare Research and AP Guidance
The theoretical debate already has a real-world counterpart. In April 2025, Anthropic announced a research program on model welfare, according to Eleos. An August 2025 Eleos post discussed allowing Claude to exit conversations.
The AP Stylebook's September 2026 guidance states that AI systems “do not think, feel, want or understand,” and advises journalists to avoid anthropomorphizing language, according to the AP Stylebook.
Lawmakers Move to Slam the Legal Door Shut
Lawmakers Consider AI Restrictions While Berkeley researchers debated early indicators of machine agency, Ohio lawmakers have also considered restrictions on AI, including Ohio House Bill 469.
The conference put a difficult question before researchers: whether AI systems could feel pain, and what obligations might follow.
