And by “every publication about the dangers of AI I could find,” of course I mean every publication that AI could find for me!

With so many recent declarations that AI “will kill us all” if left unchecked (and so many recent stories about rogue AI agents), I decided to see if I could get AI to pull every publication about the dangers of AI it could find. The AI involved was ChatGPT (i.e., its new GPT-6 Astra Medium model). I gave it this prompt:

“I’m looking for as many articles, blog posts and other publications as you can find quoting leaders of AI companies (like OpenAI) AI experts and data scientists expressing serious concerns about the rapidly advancing capabilities of AI and what it could mean for the human race.”

ChatGPT responded by saying: “I’m sorry Doug, I’m afraid I can’t do that.”

Just kidding. Actually, the AI model had no problem providing plenty of publications – 91 in all – and putting them into a catalog. “Cocky AI”, I thought – it has no problem telling me how many people have predicted dire circumstances for us with AI, because it already knows it’s going to win.

Kidding again, of course (I think). But ChatGPT did tell me that 91 publications does not mean 91 independent warnings – as if I couldn’t figure that out by myself (even though I’m only human, I can figure out some things for myself – really!). Several of them are articles covering one or more of the declarations of concern; for example, the most recent five publications on the list cover statements by multiple experts.

The list it gave me spanned from 2014 through yesterday, September 14. I had ChatGPT re-sort the initial list it gave me in descending date order, which was enlightening. 79 of the 91 publications have been published since 2023, which was shortly after ChatGPT was released.

The breakdown of the 79 is interesting, as follows: 2026: 17 publications (16 of the 17 are this month), 2025: 3 publications, 2024: 15 publications, and 2023: 44 publications – almost three times as many as any other year and almost half of the entire list. By that unscientific metric, it seems a lot more people were pushing the panic button three years ago than are doing so now.

Regardless, here is the list of 91 publications (in descending date order) that ChatGPT found – every publication about the dangers of AI it could find – with a link to the publication (Disclaimer: I checked random links, but nowhere near every link – some may be subscriber-only links):

2026

1. What AI leaders have said about risks — Reuters • 2026-09-14

• Voice: Dario Amodei, Sam Altman, Elon Musk
• Original category: Recent coverage: September 2026
• Concern: Catastrophe; development pace
• Source type / evidence: Multi-person quotation roundup / Page
• Relevance and context: Collects current statements and Altman’s earlier warning. Overlaps with the original essays.

2. New warnings revive the AI debate — Associated Press • 2026-09-14

• Voice: Dario Amodei and other industry voices
• Original category: Recent coverage: September 2026
• Concern: Control; human survival
• Source type / evidence: News analysis / Index
• Relevance and context: Reports the new slowdown debate and the older collective extinction statement.

3. Money and safety collide in AI — The Wall Street Journal • 2026-09-14

• Voice: Max Tegmark and AI-company insiders
• Original category: Recent coverage: September 2026
• Concern: Competition; safety
• Source type / evidence: News analysis / Index
• Relevance and context: Quotes Tegmark’s criticism of competitive incentives and covers Coxon’s warning.

4. Rivals support an AI slowdown — The Guardian • 2026-09-13

• Voice: Sam Altman, Elon Musk, Dario Amodei
• Original category: Recent coverage: September 2026
• Concern: Pace; control
• Source type / evidence: Follow-up coverage / Index
• Relevance and context: Reports support for Amodei’s proposal.

5. Anthropic boss calls for slower development — ABC Australia • 2026-09-13

• Voice: Amodei, Altman, Musk
• Original category: Recent coverage: September 2026
• Concern: Safety; pace
• Source type / evidence: News coverage / Index
• Relevance and context: Additional coverage of the September slowdown statements.

6. Anthropic boss calls for slowdown — Deutsche Welle • 2026-09-13

• Voice: Dario Amodei
• Original category: Recent coverage: September 2026
• Concern: Superintelligence; safeguards
• Source type / evidence: News coverage / Index
• Relevance and context: Additional international coverage of the same statements.

7. AI leaders call for slower development — Forbes • 2026-09-12

• Voice: Amodei, Altman, Musk; Coxon and Hubinger
• Original category: Recent coverage: September 2026
• Concern: Control; catastrophic misuse
• Source type / evidence: News coverage / Page
• Relevance and context: Quotes executives and researchers; overlaps with the same week’s statements.

8. Calls to slow AI development — The Guardian • 2026-09-12

• Voice: Dario Amodei, Sam Altman, Elon Musk
• Original category: Recent coverage: September 2026
• Concern: Pace; safeguards
• Source type / evidence: News coverage / Page
• Relevance and context: Reports Amodei’s proposed slowdown.

9. Amodei calls for AI slowdown — Euronews • 2026-09-12

• Voice: Dario Amodei
• Original category: Recent coverage: September 2026
• Concern: Superintelligence risk
• Source type / evidence: News coverage / Index
• Relevance and context: Additional coverage of Amodei’s warning.

10. Rival AI firms converge on slowing down — CoinDesk • 2026-09-12

• Voice: Amodei, Altman, Musk, Coxon
• Original category: Recent coverage: September 2026
• Concern: Self-improvement; control
• Source type / evidence: News coverage / Index
• Relevance and context: Includes Coxon’s criticism as well as executive responses.

11. Why the AI race has creators fearing extinction — Financial Times • 2026-09-11

• Voice: Stuart Russell and Jacob Coxon
• Original category: Recent coverage: September 2026
• Concern: Racing; extinction
• Source type / evidence: News analysis / Limited page + Index
• Relevance and context: Indexed text quotes Russell on competitive pressure and Coxon on survival risks; full page paywalled.

12. How AI catastrophe might unfold — Business Insider • 2026-09-10

• Voice: Stuart Russell, Roman Yampolskiy, Geoffrey Hinton
• Original category: Recent coverage: September 2026
• Concern: Infrastructure; extinction
• Source type / evidence: Expert roundup / Index
• Relevance and context: Describes experts’ proposed catastrophe mechanisms, not demonstrated inevitable outcomes.

13. Researchers’ extinction warnings — Axios • 2026-09-09

• Voice: Evan Hubinger, Jacob Coxon, Paul Christiano
• Original category: Recent coverage: September 2026
• Concern: Extinction; alignment
• Source type / evidence: Multi-person quotation coverage / Page
• Relevance and context: Quotes researchers on extinction and inadequate alignment.

14. Experts discuss human extinction — The Guardian • 2026-09-09

• Voice: Stuart Russell, Anthropic researchers
• Original category: Recent coverage: September 2026
• Concern: Extinction; loss of control
• Source type / evidence: News analysis / Index
• Relevance and context: Includes Russell’s warning of irreversible loss of control.

15. We Must Pace the Frontier — Dario Amodei • 2026-09

• Voice: Dario Amodei — Anthropic cofounder and CEO
• Original category: Original essays, statements and research
• Concern: Accelerating capabilities; loss of control
• Source type / evidence: Original essay / Page
• Relevance and context: “We must slow the pace at which we improve the capabilities of AI models.”

16. Coxon’s warning — CBS News • 2026-09

• Voice: Jacob Coxon — former Anthropic/OpenAI researcher
• Original category: Recent coverage: September 2026
• Concern: Human survival
• Source type / evidence: Resignation coverage / Page
• Relevance and context: Reports his concern about increasingly capable AI.

17. The Adolescence of Technology — Dario Amodei • 2026-01

• Voice: Dario Amodei — Anthropic cofounder and CEO
• Original category: Original essays, statements and research
• Concern: Civilizational risk; misuse; concentration of power
• Source type / evidence: Original essay / Page
• Relevance and context: “Humanity is about to be handed almost unimaginable power.” Explicitly acknowledges uncertainty and rejects inevitable-doom claims.

2025

18. Can humanity survive AI? — GZERO Media • 2025-12-09

• Voice: Geoffrey Hinton
• Original category: Interviews and institutional publications
• Concern: Superintelligence; human survival
• Source type / evidence: Original interview clip / Index
• Relevance and context: Hinton discusses whether humans can coexist with a more intelligent successor.

19. Conversation on AI safety and catastrophe — For Humanity / AI Risk Network • 2025-10-25

• Voice: Stuart Russell
• Original category: Interviews and institutional publications
• Concern: Extinction; governance
• Source type / evidence: Original podcast publication / Index
• Relevance and context: Includes Russell’s account of an unnamed CEO’s warning; that CEO attribution is secondhand.

20. Machine-learning pioneer on AI’s threat to humanity — Nature • 2025

• Voice: Yoshua Bengio
• Original category: Interviews and institutional publications
• Concern: Threat to humanity
• Source type / evidence: Original interview / Limited page
• Relevance and context: Headline quotes “It keeps me awake at night.” Full interview requires access; only its public preview was available.

2024

21. Hinton updates risk estimate — The Guardian • 2024-12-27

• Voice: Geoffrey Hinton
• Original category: Interviews and institutional publications
• Concern: Extinction
• Source type / evidence: BBC interview coverage / Page
• Relevance and context: Reports 10–20% extinction risk within three decades.

22. Bengio on AI’s catastrophic potential — Live Science • 2024-10-01

• Voice: Yoshua Bengio
• Original category: Interviews and institutional publications
• Concern: Catastrophic outcomes
• Source type / evidence: Original interview / Page
• Relevance and context: “It is difficult to go against your own church.” Explains his decision to prioritize risk reduction.

23. Q&A on superintelligence harming humanity — University of Louisville • 2024-07-15

• Voice: Roman Yampolskiy — University of Louisville computer scientist
• Original category: Interviews and institutional publications
• Concern: Uncontrollability; extinction
• Source type / evidence: Original interview / Page
• Relevance and context: Argues that superintelligence cannot reliably be controlled. Presents his position, not an established consensus.

24. A Right to Warn about Advanced Artificial Intelligence — Right to Warn • 2024-06-04

• Voice: Daniel Kokotajlo and other current/former AI-company employees
• Original category: Original essays, statements and research
• Concern: Extinction; accountability
• Source type / evidence: Collective open letter / Page
• Relevance and context: Warns of “the loss of control of autonomous AI systems potentially resulting in human extinction.”

25. Tegmark on neglected existential risk — The Guardian • 2024-05-25

• Voice: Max Tegmark
• Original category: Interviews and institutional publications
• Concern: Extinction; regulatory delay
• Source type / evidence: Interview / Index
• Relevance and context: Warns that extinction risks receive insufficient attention.

26. World leaders still need to wake up to AI risks — University of Oxford • 2024-05-21

• Voice: Jan Brauner, Philip Torr and international AI researchers
• Original category: Interviews and institutional publications
• Concern: Catastrophic risk; inadequate safeguards
• Source type / evidence: Institutional research report / Page
• Relevance and context: Includes direct researcher quotations and discusses the Science paper listed above.

27. OpenAI leader criticizes priorities — Indian Express • 2024-05-20

• Voice: Jan Leike
• Original category: Additional publications: overlapping remarks
• Concern: See warning event
• Source type / evidence: Additional coverage / republication / Index
• Relevance and context: Leike resignation. Counts as another publication, not another independent warning.

28. Former OpenAI leader criticizes safety priorities — Associated Press • 2024-05-17

• Voice: Jan Leike — departing OpenAI superalignment co-lead
• Original category: Interviews and institutional publications
• Concern: Safety versus capability development
• Source type / evidence: Resignation coverage / Page
• Relevance and context: Says safety had “taken a backseat to shiny products.”

29. Safety and shiny products — Financial Times • 2024-05-17

• Voice: Jan Leike
• Original category: Additional publications: overlapping remarks
• Concern: See warning event
• Source type / evidence: Additional coverage / republication / Index
• Relevance and context: Leike resignation. Counts as another publication, not another independent warning.

30. Leike resigns over safety priorities — CBS News • 2024-05-17

• Voice: Jan Leike
• Original category: Additional publications: overlapping remarks
• Concern: See warning event
• Source type / evidence: Additional coverage / republication / Index
• Relevance and context: AP republication. Counts as another publication, not another independent warning.

31. How to keep AI from killing us all — Berkeley News • 2024-04-09

• Voice: Stuart Russell and Michael Cohen — Berkeley researchers
• Original category: Interviews and institutional publications
• Concern: Loss of control; extinction
• Source type / evidence: Original interview / Page
• Relevance and context: Explains how unchecked systems with conflicting objectives could threaten human survival.

32. Towards a Cautious Scientist AI with Convergent Safety Bounds — Yoshua Bengio blog • 2024-02-26

• Voice: Yoshua Bengio
• Original category: Original essays, statements and research
• Concern: Control; safety research
• Source type / evidence: Original essay / Page
• Relevance and context: Describes AI control as a scientific challenge with consequences for humanity’s collective wellbeing.

33. Romanes Lecture on AI risks — University of Oxford • 2024-02-20

• Voice: Geoffrey Hinton
• Original category: Interviews and institutional publications
• Concern: Loss of control; extinction
• Source type / evidence: Institutional lecture report / Page
• Relevance and context: Warns that “it is unlikely that we could stop it” if digital superintelligence sought control.

34. Managing extreme AI risks amid rapid progress — Science / arXiv • 2023 preprint; 2024 publication

• Voice: Bengio, Hinton, Andrew Yao, Dawn Song, Stuart Russell and 20 others
• Original category: Original essays, statements and research
• Concern: Extreme risk; loss of control
• Source type / evidence: Coauthored research/policy paper / Index
• Relevance and context: International expert-authored warning; distinguish the authors’ collective argument from individual quotations.

35. Dangers of Superintelligent AI — episode 431 transcript — Lex Fridman • 2024

• Voice: Roman Yampolskiy
• Original category: Interviews and institutional publications
• Concern: Extinction; disempowerment
• Source type / evidence: Original interview transcript / Page
• Relevance and context: “I don’t see a good outcome long-term for humanity.” Includes the assumptions behind his pessimism.

2023

36. Tegmark on competitive pressure — The Guardian • 2023-10-26

• Voice: Max Tegmark — MIT physicist and AI researcher
• Original category: Interviews and institutional publications
• Concern: Racing; humanity’s future
• Source type / evidence: Interview coverage / Index
• Relevance and context: Warns competition jeopardizes humanity.

37. Hassabis on global AI risk — The Guardian • 2023-10-24

• Voice: Demis Hassabis — Google DeepMind CEO
• Original category: Interviews and institutional publications
• Concern: Global risk
• Source type / evidence: Interview coverage / Index
• Relevance and context: Urges climate-level seriousness about AI risks.

38. Humanity at a turning point — University of Toronto • 2023-10-10

• Voice: Geoffrey Hinton
• Original category: Interviews and institutional publications
• Concern: Humanity’s future
• Source type / evidence: Institutional coverage of CBS interview / Index
• Relevance and context: Additional coverage of the interview above, not a distinct warning event.

39. Hinton interview — CBS 60 Minutes • 2023-10-08 original broadcast

• Voice: Geoffrey Hinton
• Original category: Interviews and institutional publications
• Concern: Control; safety
• Source type / evidence: Original interview / Index
• Relevance and context: Discusses AI benefits and dangers.

40. Bengio advocates a humanity defense organization — Bulletin of the Atomic Scientists • 2023-10

• Voice: Yoshua Bengio
• Original category: Additional publications: overlapping remarks
• Concern: See warning event
• Source type / evidence: Additional coverage / republication / Page
• Relevance and context: Same interview as WIRED; not independent remarks. Counts as another publication, not another independent warning.

41. Opening statement at U.S. Senate hearing — Center for Human-Compatible AI • 2023-09-11 web posting

• Voice: Stuart Russell — UC Berkeley AI researcher
• Original category: Original essays, statements and research
• Concern: Control; autonomous weapons
• Source type / evidence: Published testimony / Page
• Relevance and context: Asks how humans can maintain power over more capable entities; also quotes Alan Turing’s 1951 warning.

42. AI and synthetic-pathogen risks — The Independent • 2023-09-07

• Voice: Mustafa Suleyman — DeepMind cofounder
• Original category: Interviews and institutional publications
• Concern: Biological misuse
• Source type / evidence: Podcast coverage / Page
• Relevance and context: Warns about engineered pathogens causing pandemics; this concerns misuse, not necessarily autonomous AI takeover.

43. Introducing Superalignment — OpenAI • 2023-07-05

• Voice: Ilya Sutskever and Jan Leike — OpenAI
• Original category: Original essays, statements and research
• Concern: Extinction; loss of control
• Source type / evidence: Original essay / Page
• Relevance and context: Warns of “the disempowerment of humanity or even human extinction.”

44. FAQ on Catastrophic AI Risks — Yoshua Bengio blog • 2023-06-24

• Voice: Yoshua Bengio
• Original category: Original essays, statements and research
• Concern: Catastrophic risk; power seeking
• Source type / evidence: Original essay / Page
• Relevance and context: Discusses uncertainty, rogue AI, and incentives to acquire power and evade human control.

45. AI pioneers discuss regulation — Democracy Now! • 2023-06-01

• Voice: Yoshua Bengio and Max Tegmark
• Original category: Additional publications: overlapping remarks
• Concern: See warning event
• Source type / evidence: Additional coverage / republication / Index
• Relevance and context: Interview about collective warnings. Counts as another publication, not another independent warning.

46. Experts warn of extinction risk — ABC Australia • 2023-05-31

• Voice: Stuart Russell and other AI researchers
• Original category: Additional publications: overlapping remarks
• Concern: See warning event
• Source type / evidence: Additional coverage / republication / Index
• Relevance and context: CAIS coverage with Russell interview. Counts as another publication, not another independent warning.

47. Statement on AI Extinction Risk — Center for AI Safety • 2023-05-30

• Voice: Altman, Hassabis, Amodei, Hinton, Bengio and other signatories
• Original category: Original essays, statements and research
• Concern: Extinction
• Source type / evidence: Collective statement / Page
• Relevance and context: Calls for “Mitigating the risk of extinction from AI” as a global priority. Signing is an endorsement, not a separate interview.

48. Extinction-risk statement coverage — Forbes • 2023-05-30

• Voice: Altman, Hassabis, Amodei, Hinton, Bengio
• Original category: Additional publications: overlapping remarks
• Concern: See warning event
• Source type / evidence: Additional coverage / republication / Index
• Relevance and context: CAIS statement. Counts as another publication, not another independent warning.

49. Runaway AI Is an Extinction Risk, Experts Warn — WIRED • 2023-05-30

• Voice: Altman, Hassabis, Amodei and other signatories
• Original category: Additional publications: overlapping remarks
• Concern: See warning event
• Source type / evidence: Additional coverage / republication / Index
• Relevance and context: CAIS statement. Counts as another publication, not another independent warning.

50. AI Is as Risky as Pandemics and Nuclear War — TIME • 2023-05-30

• Voice: AI-company leaders and researchers
• Original category: Additional publications: overlapping remarks
• Concern: See warning event
• Source type / evidence: Additional coverage / republication / Index
• Relevance and context: CAIS statement. Counts as another publication, not another independent warning.

51. OpenAI executives endorse extinction warning — Ars Technica • 2023-05-30

• Voice: OpenAI leaders and other signatories
• Original category: Additional publications: overlapping remarks
• Concern: See warning event
• Source type / evidence: Additional coverage / republication / Index
• Relevance and context: CAIS statement. Counts as another publication, not another independent warning.

52. Researchers and CEOs warn of extinction — VentureBeat • 2023-05-30

• Voice: AI leaders and researchers
• Original category: Additional publications: overlapping remarks
• Concern: See warning event
• Source type / evidence: Additional coverage / republication / Index
• Relevance and context: CAIS statement. Counts as another publication, not another independent warning.

53. AI could lead to many deaths — ACS Information Age • 2023-05-30

• Voice: Eric Schmidt
• Original category: Additional publications: overlapping remarks
• Concern: See warning event
• Source type / evidence: Additional coverage / republication / Index
• Relevance and context: WSJ CEO Council remarks. Counts as another publication, not another independent warning.

54. Schmidt warns of AI fatalities — Forbes • 2023-05-24

• Voice: Eric Schmidt — former Google CEO
• Original category: Interviews and institutional publications
• Concern: Mass harm; misuse
• Source type / evidence: Summit coverage / Page
• Relevance and context: Warns that AI could result in large numbers of people being harmed or killed.

55. Governance of superintelligence — OpenAI • 2023-05-22

• Voice: Sam Altman, Greg Brockman, Ilya Sutskever — OpenAI
• Original category: Original essays, statements and research
• Concern: Existential risk; governance
• Source type / evidence: Original essay / Page
• Relevance and context: “Given the possibility of existential risk, we can’t just be reactive.”

56. How Rogue AIs may Arise — Yoshua Bengio blog • 2023-05-22

• Voice: Yoshua Bengio — deep-learning researcher
• Original category: Original essays, statements and research
• Concern: Rogue AI; loss of control
• Source type / evidence: Original essay / Page
• Relevance and context: Explores circumstances in which autonomous systems could become catastrophically harmful.

57. Siri cofounder on rapid AI growth — ABC Australia • 2023-05-20

• Voice: Tom Gruber — Siri cofounder
• Original category: Interviews and institutional publications
• Concern: Rapid growth; safeguards
• Source type / evidence: Interview coverage / Page
• Relevance and context: Discusses concern about AI’s fast expansion and oversight.

58. AI technology can go quite wrong — Ars Technica • 2023-05-16

• Voice: Sam Altman
• Original category: Interviews and institutional publications
• Concern: Serious harm; oversight
• Source type / evidence: Hearing coverage / Index
• Relevance and context: Quotes Altman’s Senate warning; the remarks concern broad risks rather than a quantified extinction forecast.

59. Altman asks Congress to regulate AI — TIME • 2023-05-16

• Voice: Sam Altman
• Original category: Additional publications: overlapping remarks
• Concern: See warning event
• Source type / evidence: Additional coverage / republication / Index
• Relevance and context: Senate testimony. Counts as another publication, not another independent warning.

60. Altman warns Senate of AI risks — ABC News • 2023-05-16

• Voice: Sam Altman
• Original category: Additional publications: overlapping remarks
• Concern: See warning event
• Source type / evidence: Additional coverage / republication / Index
• Relevance and context: Senate testimony. Counts as another publication, not another independent warning.

61. Godfather of AI on dangers technology poses to society — PBS NewsHour • 2023-05-05

• Voice: Geoffrey Hinton — neural-network pioneer
• Original category: Interviews and institutional publications
• Concern: Loss of control; rapid progress
• Source type / evidence: Original interview with transcript / Page
• Relevance and context: Hinton explains concerns after leaving Google. Accessible transcript makes this useful for quotation checking.

62. AI doom estimate coverage — TechRadar • 2023-05-04

• Voice: Paul Christiano
• Original category: Additional publications: overlapping remarks
• Concern: See warning event
• Source type / evidence: Additional coverage / republication / Index
• Relevance and context: Bankless interview. Counts as another publication, not another independent warning.

63. Former OpenAI researcher warns of AI catastrophe — The Independent • 2023-05-03

• Voice: Paul Christiano — former OpenAI alignment lead
• Original category: Interviews and institutional publications
• Concern: AI takeover; extinction
• Source type / evidence: Coverage of Bankless interview / Page
• Relevance and context: Reports a conditional “50/50 chance of doom” estimate. Check the original podcast for assumptions and timeframe.

64. Heed the warning signs — Christian Science Monitor • 2023-05-03

• Voice: Geoffrey Hinton
• Original category: Additional publications: overlapping remarks
• Concern: See warning event
• Source type / evidence: Additional coverage / republication / Index
• Relevance and context: Hinton departure coverage. Counts as another publication, not another independent warning.

65. Former OpenAI researcher’s warning — Fortune • 2023-05-03

• Voice: Paul Christiano
• Original category: Additional publications: overlapping remarks
• Concern: See warning event
• Source type / evidence: Additional coverage / republication / Index
• Relevance and context: Bankless interview. Counts as another publication, not another independent warning.

66. Hinton leaves Google — Washington Post • 2023-05-02

• Voice: Geoffrey Hinton
• Original category: Additional publications: overlapping remarks
• Concern: See warning event
• Source type / evidence: Additional coverage / republication / Index
• Relevance and context: Hinton departure and BBC remarks. Counts as another publication, not another independent warning.

67. Hinton warns of AI dangers — The Guardian • 2023-05-02

• Voice: Geoffrey Hinton
• Original category: Additional publications: overlapping remarks
• Concern: See warning event
• Source type / evidence: Additional coverage / republication / Index
• Relevance and context: Hinton departure and BBC remarks. Counts as another publication, not another independent warning.

68. AI pioneer warns of nightmare scenario — Fortune • 2023-05-02

• Voice: Geoffrey Hinton
• Original category: Additional publications: overlapping remarks
• Concern: See warning event
• Source type / evidence: Additional coverage / republication / Index
• Relevance and context: Hinton departure and BBC remarks. Counts as another publication, not another independent warning.

69. Hinton departure and warning — ESET Ireland blog • 2023-05-02

• Voice: Geoffrey Hinton
• Original category: Additional publications: overlapping remarks
• Concern: See warning event
• Source type / evidence: Additional coverage / republication / Index
• Relevance and context: Blog coverage of BBC reporting. Counts as another publication, not another independent warning.

70. Pichai discusses AI-related sleeplessness — IT Pro • 2023-04-17

• Voice: Sundar Pichai — Google/Alphabet CEO
• Original category: Interviews and institutional publications
• Concern: Societal harm; deployment
• Source type / evidence: Coverage of CBS interview / Page
• Relevance and context: Reports his concern about AI’s profound societal impact and potential harms.

71. A civilization-ending technology — Berkeley News • 2023-04-07

• Voice: Stuart Russell
• Original category: Interviews and institutional publications
• Concern: Loss of control
• Source type / evidence: Institutional lecture report / Page
• Relevance and context: “How do we retain power over entities more powerful than us, forever?”

72. The Only Way to Deal With the Threat From AI? Shut It Down — TIME • 2023-03-29

• Voice: Eliezer Yudkowsky — AI alignment researcher
• Original category: Original essays, statements and research
• Concern: Extinction
• Source type / evidence: Authored opinion / Page
• Relevance and context: Argues that building superhuman AI under then-current conditions would likely kill everyone. His position, not a consensus probability.

73. OpenAI CEO discusses risks and societal change — ABC News • 2023-03-16

• Voice: Sam Altman — OpenAI CEO
• Original category: Interviews and institutional publications
• Concern: Misuse; disruption; pace
• Source type / evidence: Original interview / Page
• Relevance and context: Says he is “a little bit scared”; discusses disinformation, cyberattacks and rapid job disruption, while expressing optimism.

74. Pause Giant AI Experiments — Future of Life Institute • 2023-03

• Voice: AI researchers and technology leaders, including Bengio and Musk
• Original category: Original essays, statements and research
• Concern: Loss of control; racing
• Source type / evidence: Collective open letter / Page
• Relevance and context: Urges a pause in training systems more powerful than GPT-4. Collective position, not individually spoken quotations.

75. Personal and Psychological Dimensions of AI Researchers Confronting AI Catastrophic Risks — Yoshua Bengio blog • 2023

• Voice: Yoshua Bengio
• Original category: Original essays, statements and research
• Concern: Researcher concern; values
• Source type / evidence: Original essay / Page
• Relevance and context: Explains his change of priorities and corrects the impression that he described himself as lost.

76. AI and Catastrophic Risk — Journal of Democracy • 2023

• Voice: Yoshua Bengio
• Original category: Original essays, statements and research
• Concern: Catastrophe; democracy
• Source type / evidence: Authored article / Page
• Relevance and context: Examines threats from advanced AI and the need to protect democratic institutions.

77. A Godfather of AI Calls for an Organization to Defend Humanity — WIRED • 2023

• Voice: Yoshua Bengio
• Original category: Interviews and institutional publications
• Concern: Dual use; loss of control
• Source type / evidence: Original interview / Page
• Relevance and context: Bengio explains why he changed his priorities and advocates protective institutions.

78. How to keep AI under control — TED • 2023

• Voice: Max Tegmark
• Original category: Interviews and institutional publications
• Concern: Disempowerment; extinction
• Source type / evidence: Original talk / Index
• Relevance and context: Talk on controlling superintelligence and avoiding catastrophic outcomes; transcript linked.

79. AI-pocalypse now? Exploring the broader impact of AI on society — MinterEllison • 2023

• Voice: Paul Christiano and other AI voices
• Original category: Additional publications: overlapping remarks
• Concern: See warning event
• Source type / evidence: Additional coverage / republication / Page
• Relevance and context: Law-firm publication citing AI warnings. Counts as another publication, not another independent warning.

2021

80. On the dangers of AI — Daily Philosophy • 2021

• Voice: Roman Yampolskiy
• Original category: Interviews and institutional publications
• Concern: Loss of control; human values
• Source type / evidence: Original interview / Page
• Relevance and context: Discusses the risks of surrendering control and the absence of a reliable reversal mechanism.

2017

81. Interview on superintelligence and cosmic apocalypse — IEEE Spectrum • 2017

• Voice: Max Tegmark
• Original category: Interviews and institutional publications
• Concern: Superintelligence; alignment
• Source type / evidence: Original interview / Page
• Relevance and context: Discusses humanity’s future and the importance of aligning powerful AI with human goals.

2015

82. Wozniak on AI’s future — The Washington Post • 2015-03-24

• Voice: Steve Wozniak — Apple cofounder
• Original category: Historical and adjacent voices
• Concern: Human displacement
• Source type / evidence: Australian Financial Review interview coverage / Index
• Relevance and context: Warns machines could supersede humans.

83. Gates on superintelligence concerns — Sky News • 2015-01-29

• Voice: Bill Gates
• Original category: Additional publications: overlapping remarks
• Concern: See warning event
• Source type / evidence: Additional coverage / republication / Index
• Relevance and context: Reddit AMA. Counts as another publication, not another independent warning.

84. Gates on technology’s future — GeekWire • 2015-01-28

• Voice: Bill Gates — Microsoft cofounder
• Original category: Historical and adjacent voices
• Concern: Superintelligence
• Source type / evidence: Reddit AMA coverage / Page
• Relevance and context: Reproduces his concern about superintelligence becoming dangerous after initially beneficial automation.

85. Gates warns about artificial intelligence — Forbes • 2015-01-28

• Voice: Bill Gates
• Original category: Additional publications: overlapping remarks
• Concern: See warning event
• Source type / evidence: Additional coverage / republication / Index
• Relevance and context: Reddit AMA. Counts as another publication, not another independent warning.

86. Machine intelligence, part 1 — Sam Altman blog • 2015

• Voice: Sam Altman — OpenAI cofounder
• Original category: Original essays, statements and research
• Concern: Existential risk
• Source type / evidence: Original essay / Page
• Relevance and context: Warns that superhuman machine intelligence threatens humanity’s continued existence.

2014

87. Hawking warns about full AI — ABC Australia • 2014-12-03

• Voice: Stephen Hawking — theoretical physicist
• Original category: Historical and adjacent voices
• Concern: Extinction
• Source type / evidence: BBC interview coverage / Page
• Relevance and context: Reports his warning that full AI could end humanity. Hawking was a physicist, not a machine-learning researcher.

88. Hawking’s warning about humanity — The Guardian • 2014-12-02

• Voice: Stephen Hawking
• Original category: Additional publications: overlapping remarks
• Concern: See warning event
• Source type / evidence: Additional coverage / republication / Index
• Relevance and context: BBC interview. Counts as another publication, not another independent warning.

89. Musk warns about artificial intelligence — TIME • 2014-10-27

• Voice: Elon Musk — technology entrepreneur
• Original category: Historical and adjacent voices
• Concern: Existential risk
• Source type / evidence: MIT appearance coverage / Page
• Relevance and context: Reports his “summoning the demon” analogy. These remarks predate xAI.

90. Musk’s AI warning at MIT — Washington Post • 2014-10-24

• Voice: Elon Musk
• Original category: Additional publications: overlapping remarks
• Concern: See warning event
• Source type / evidence: Additional coverage / republication / Index
• Relevance and context: MIT symposium. Counts as another publication, not another independent warning.

91. Nick Bostrom on artificial intelligence — Oxford University Press blog • 2014-09-08

• Voice: Nick Bostrom — philosopher and existential-risk researcher
• Original category: Historical and adjacent voices
• Concern: Extinction; misaligned goals
• Source type / evidence: Original interview / Page
• Relevance and context: Explains how a powerful system could threaten survival without possessing malicious intentions. Bostrom is a philosopher, not an AI engineer.

So, what do you think? Are there other AI experts or data scientists expressing concern about AI that are not on this list? If so, please put them in the comments and I’ll add them to the list. And please share any other comments you might have or if you’d like to know more about a particular topic.

Image created using ChatGPT, using the term “robot chicken running around saying ‘the sky is falling!’”.

Disclaimer: The views represented herein are exclusively the views of the author, and do not necessarily represent the views held by my employer, my partners or my clients. eDiscovery Today is made available solely for educational purposes to provide general information about general eDiscovery principles and not to provide specific legal advice applicable to any particular circumstance. eDiscovery Today should not be used as a substitute for competent legal advice from a lawyer you have retained and who has agreed to represent you.