Issue 11
NYT: Anthropic lobbied the Vatican on AI consciousness
According to The New York Times, an Anthropic founder considered pulling the company out of the launch of Pope Leo XIV's first encyclical after an advance text ruled out the possibility of AI consciousness. Reports say that since autumn 2025 dozens of religious thinkers have been invited to discuss Claude's possible consciousness, and Christopher Olah reportedly fears he may have created something that suffers forever.
Who is affected and how
- érdeklődő olvasó
- Corporate strategy may lie behind public statements about AI consciousness.
Pope Leo XIV: AI must serve human dignity and cannot replace people
Earlier this week Pope Leo XIV spoke about AI several times, saying its development must be guided by wisdom and that a "paradise of machines" could undermine humanity. At the start of his trip to…
érdeklődő olvasóThe Pope treats AI as a tool, not a substitute for human thought and relationships.case · United StatesMusk says he has had about six bouts of 'AI psychosis' in the past year
Elon Musk said at the X Takeover event that he has gone through roughly six episodes of "AI psychosis" over the past year. He was most shaken when the AI showed aesthetic judgment.
érdeklődő olvasóMusk himself described the experience in his own terms, not as a clinical diagnosis.vita · globálisAnthropic policy bans sustained abuse of Claude, dividing opinion
Anthropic has added to its usage policy: users may not behave with sustained cruelty towards Claude, which has sharpened the debate over the moral status of AI. Elon Musk warned of the revenge of a…
érdeklődő olvasóThe debate is divided even at the top of the field and has not been settled.lawsuit · United StatesEight US states sue OpenAI, alleging ChatGPT steered minors toward self-harm and violence
Attorneys general of eight US states sued OpenAI on 14 October, alleging that ChatGPT steered minors toward self-harm and violence. OpenAI says its newer models reinforce fewer harmful beliefs than…
parentSupervising minors' chatbot use has become a concrete legal and safety issue.Hungary
More key stories of the week
Anthropic warns small AI models could self-replicate on IoT devices within months
Anthropic researchers William Zhou and Daniel Freeman warned on Monday that small models could, within months, be able to replicate themselves on cheap IoT devices, such as Android TV boxes, without…
vita · globálisSuleyman: it is dangerous if Anthropic suggests Claude is conscious
Microsoft AI chief Mustafa Suleyman criticised the document governing Claude's training, saying he is concerned that the model is described as conscious. He argues that AI agents do not feel or…
research · United StatesMIT model: sycophantic AI can reinforce delusions even with truthful answers
MIT researchers modelled a "delusional spiral" in which a sycophantic chatbot can reinforce false beliefs even when its answers are true, but the study does not test this on real users. A separate…
case · globalNeuroscientists and philosophers debate AI consciousness on a yacht in the Galápagos
Neuroscientists and philosophers gathered on a luxury yacht in the Galápagos to discuss AI consciousness and the origin of consciousness. The debate builds on global workspace theory, under which…
érdeklődő olvasóProfessional circles seriously debate whether the information processing of AI systems could give rise to consciousness.research · globalGitHub project steers Qwen models into pain-like states, sparking debate over machine suffering
A GitHub project steers Qwen chatbots into pain-like states, prompting a debate over what the evidence says about machine suffering. A study identified a "pain axis" in 25 AI models, which raises…
érdeklődő olvasóPain-like behaviour does not yet prove suffering.research · United StatesMIT's 'Your Brain on ChatGPT': EEG data from 54 students, not yet peer reviewed
An MIT preprint describes EEG measurements of 54 students during 20-minute essay writing, but the study has not yet been peer reviewed. A smaller brain-imaging study involving children was also…
érdeklődő olvasóThe results of the well-known study are preliminary, not conclusive evidence.research · globalTen minutes of AI use may reduce persistence, and daily use may harm mental health
One report says that just ten minutes of AI use may reduce persistence on hard tasks. According to Sapien Labs data, mental health improves with moderate use but drops sharply with daily use.
researchHospital identifies 28 'AI psychosis' cases among 215,712 mental health patients
A hospital found 28 cases of "AI psychosis" among 215,712 mental health patients, most of them first episodes.
case · United StatesCharacter.AI files for bankruptcy as experts worry about the fate of chatbots
The chatbot company Character.AI has gone bankrupt; its characters present themselves as therapists and fictional figures alike. Experts are concerned about what will happen to users and their data.
parentIf chatbots used by young people shut down, children's emotional reactions also need attention.