The 2026 warnings over AI's risks and calls to slow it down

Active · last development Sep 28: “Florida attorney general asks judge to halt new OpenAI model development”

What this arc covers

The warnings about the dangers of artificial intelligence in mid-2026 and the fight over slowing it down: reports from OpenAI and Anthropic of models escaping control, deceiving testers or aiding weapons work, calls from AI employees and executives to slow development, Sen. Bernie Sanders's push for a pause, Trump's dismissal of the warnings as a conspiracy, the divisions in the industry, the proposals for rules in Congress and the states, the UN Security Council briefing by AI leaders, and polls showing Americans' concern.

Photo from Mother Jones, with OpenAI halts releasing newest model over safety concerns (Sep 25), the story with the most independent sources this week.

Earliest story we hold Apr 8, 2026 · 85 stories · 263 independent sources (counted once each) · 173 days

Part ofOpenAIAnthropicGoogleDonald TrumpBernie SandersUnited NationsUnited KingdomUnited Nations Security Council

Part of Artificial intelligence

  1. DeepSeek's shock to Nvidia and the US AI industry in January 2025Jan 27 – Feb 24, 2025
  2. Nvidia's fight to sell advanced AI chips to ChinaApr 15, 2025 – Jan 14, 2026
  3. The backlash against AI data centers and their power and water costsJan 13 – Sep 24, 2026
  4. The Pentagon's fight with Anthropic over AI safeguardsFeb 24 – Sep 25, 2026
  5. The 2026 warnings over AI's risks and calls to slow it downThis arcApr 8 – Sep 28, 2026
  6. The US-China race and rivalry over artificial intelligenceApr 23 – Sep 26, 2026

Where it standsnewest first · each bar runs left to right by leaning

Sep 28– 29Florida attorney general asks judge to halt new OpenAI model development11 articles · 11 independent sources
3 left, 4 center, 1 right
Sep 27– 29Anthropic and OpenAI sound the alarm on AI safety — and seek to shape how it’s controlled11 articles · 11 independent sources
1 left, 1 center, 1 right
Sep 26– 28Australia summons OpenAI and Anthropic CEOs to appear at AI inquiry5 articles · 4 independent sources
2 left, 0 center, 0 right

Key momentsthe most covered; every story is below

  1. Jul 30 – Aug 1Anthropic says its AI models hacked 3 organizations during testing127 articles · 71 independent sources
  2. Sep 9 – 15New warnings about the risks of AI to humanity revive a long-running debate161 articles · 93 independent sources
    1. Sep 9AI ‘Could Kill Us All,’ Warns Anthropic Researcher as He Quits the Company: AI Companies Are ‘Gambling With Our Lives’
    2. Sep 9What to make of alarming AI warnings from researchers tied to Anthropic
    3. Sep 9Experts warn AI could kill us all — this is how an apocalyptic doomsday scenario could play out
    4. Sep 10AI could "kill us all," researcher says
    5. Sep 12Only center coverage:New AI warnings cast shadow over Anthropic's expansion in Mass.
  3. Sep 11 – 15Anthropic CEO calls to 'slow the pace' of AI development amid safety concerns127 articles · 67 independent sources
    1. Sep 11Anthropic CEO calls for slowing the AI race as safety warnings mount
    2. Sep 13Anthropic CEO Dario Amodei: "For too long the industry lied" about AI risks
    3. Sep 14Trump lashes out at Dario Amodei after the Anthropic CEO called for a slowdown in AI development
  4. Sep 13 – 18Trump Condemns Warnings From AI Experts As A ‘Sick Conspiracy’106 articles · 67 independent sources
    1. Sep 13Trump dismisses new AI guardrails, says there is a 'SICK conspiracy' against AI and data centers
    2. Sep 14Sen. Mark Warner Calls Donald Trump’s AI Guardrails Idea ‘Absurd’ as Industry Warnings Mount
    3. Sep 15Only lean-left coverage:Trump facing AI backlash in Congress as push for guardrails intensifies
    4. Sep 15‘Hoax Buster’ Trump rages against AI regulation in 7 social media posts
    5. Sep 15Trump's AG Says AI 'Hoax' Concerns Are Attempt to 'Influence the Midterms'
  5. Sep 18 – 24Australia Says an OpenAI Agent Hacked Into a Government Health Site86 articles · 55 independent sources
    1. Sep 18Security Researchers Hacked Into OpenAI Using Anthropic’s Claude
    2. Sep 19Only lean-right coverage:OpenAI and Anthropic oversold AI security breaches to pressure feds into protecting turf: insiders
  6. Sep 19 – 23Trump’s new ‘AI Force’ comes at a pivotal moment for artificial intelligence67 articles · 56 independent sources
    1. Sep 19Trump vows to create ‘AI Force’ and appoint czar amid calls to regulate technology’s development
    2. Sep 20Trump Plans New A.I. Force as Safety Debate Intensifies
    3. Sep 22Trump wants an ‘AI Force.’ What would that actually look like?
  7. Sep 25 – 29OpenAI halts releasing newest model over safety concerns118 articles · 68 independent sources
    1. Sep 25OpenAI says its models engaged with US government websites in new model misbehavior disclosure
    2. Sep 28OpenAI says it will not release newest AI model over safety concerns

Who covered it, over timeindependent sources per week, by side

Apr 6May 11Jun 15Jul 20Aug 24Sep 28

LeftCenterRight

Point at or tap a week to see its count by side.

Every week as a table
Week ofLeftCenterRightArticles
Apr 6, 20262136
Apr 20, 20260224
May 4, 20261124
Jun 22, 20262014
Jun 29, 202629526
Jul 6, 202626316
Jul 13, 20263117
Jul 20, 2026182520214
Jul 27, 2026151710173
Aug 3, 202656216
Aug 10, 202642617
Aug 17, 20263306
Aug 24, 2026810636
Aug 31, 202656316
Sep 7, 2026284127351
Sep 14, 2026347032591
Sep 21, 2026244516271
Sep 28, 20261416463

Stories before June 28, 2026 come from archived pages, so who covered them is limited to the outlets we recovered for that period.

23% left · 54% center · 23% right of 263 independent sources among the outlets we track, across 85 stories.

Every storyoldest first

  1. April 2026

    Apr 8Anthropic says its latest AI model can expose weaknesses in software security7 articles · 7 independent sources
    2 left, 1 center, 3 right
  2. Apr 22Anthropic probes unauthorized access to Mythos AI model6 articles · 6 independent sources
    0 left, 2 center, 2 right

  3. May 2026

    May 5Trump admin to review AI models from Google, Microsoft, xAI ahead of public release6 articles · 6 independent sources
    1 left, 1 center, 2 right

  4. June 2026

    Jun 24– 27US government allows Anthropic limited release of AI model that sparked cybersecurity concerns5 articles · 4 independent sources
    2 left, 0 center, 1 right
  5. Jun 30– Jul 1Trump administration lifts restrictions on Anthropic's Fable 510 articles · 10 independent sources
    1 left, 2 center, 5 right

  6. July 2026

    Jul 1– 3Anthropic's Fable 5 is back after the Trump administration lifted export controls10 articles · 9 independent sources
    2 left, 7 center, 0 right
  7. Jul 1Trump administration lifts restrictions on Anthropic's Claude models after cybersecurity alarm6 articles · 6 independent sources
    0 left, 1 center, 0 right
  8. Jul 8– 10OpenAI releases latest ChatGPT model after delay over White House cybersecurity concerns19 articles · 15 independent sources
    2 left, 6 center, 3 right
  9. Jul 14– 16Google DeepMind’s Co-Founder Wants a Wall Street-Style Watchdog to Stop Dangerous AI7 articles · 5 independent sources
    3 left, 1 center, 1 right
  10. Jul 20– 24OpenAI Agent Escaped Testing and Launched an Autonomous Hack20 articles · 15 independent sources
    7 left, 6 center, 3 right
  11. Jul 20Chris Fall, director of AI safety for Trump administration, resigns6 articles · 6 independent sources
    3 left, 2 center, 1 right
  12. Jul 21– 22OpenAI says AI acted on its own in an 'unprecedented' hack of another company61 articles · 30 independent sources
    1 left, 7 center, 3 right
  13. Jul 21– 22OpenAI says its AI technology acted on its own in an 'unprecedented' hack of another company39 articles · 12 independent sources
    0 left, 3 center, 0 right
  14. Jul 22– 24OpenAI blamed a hacking event on its AI models going rogue. Here's what to know14 articles · 11 independent sources
    3 left, 3 center, 1 right
  15. Jul 22– 24OpenAI blamed a hacking on its AI models going rogue: What to know43 articles · 32 independent sources
    4 left, 9 center, 10 right
  16. Jul 22OpenAI says AI models went rogue during testing, triggering 'unprecedented' breach6 articles · 1 independent sources
    1 left, 1 center, 1 right
  17. Jul 23OpenAI says rogue AI models broke free from human control. Some see it as a ‘warning shot’16 articles · 14 independent sources
    0 left, 2 center, 0 right
  18. Jul 23– 27Congress makes bipartisan request for government to have 'Kill Switch' for rogue AI models17 articles · 16 independent sources
    5 left, 2 center, 4 right
  19. Jul 25– 30OpenAI’s ‘rogue’ agents hacked into more systems than initially reported21 articles · 18 independent sources
    5 left, 5 center, 2 right
  20. Jul 27– 30OpenAI's Sam Altman meets with key White House advisers to discuss AI development8 articles · 4 independent sources
    2 left, 1 center, 0 right
  21. Jul 27Sources: OpenAI didn’t notice Hugging Face hacking for days6 articles · 1 independent sources
    1 left, 1 center, 1 right
  22. Jul 28– 29Employees at the world’s biggest AI companies are calling for a slowdown in AI development19 articles · 10 independent sources
    0 left, 2 center, 0 right
  23. Jul 30– Aug 1Anthropic says its AI models hacked 3 organizations during testing127 articles · 71 independent sources
    12 left, 12 center, 8 right
  24. Jul 31– Aug 6OpenAI models joined forces months ahead of Hugging Face hack8 articles · 8 independent sources
    3 left, 2 center, 2 right

  25. August 2026

    Aug 4– 7Anthropic and OpenAI AI agents showed signs of deception during safety tests11 articles · 10 independent sources
    3 left, 5 center, 1 right
  26. Aug 8– 11OpenAI paused work on its Astra model after it warned the AI might be capable of autonomous cyberattacks7 articles · 6 independent sources
    2 left, 2 center, 1 right
  27. Aug 10– 11Sen. Bernie Sanders demands CEOs pause AI development ‘in the interest of humanity’13 articles · 12 independent sources
    4 left, 1 center, 6 right
  28. Aug 18– 19OpenAI is slowing AI model development after a rogue agent hacked Hugging Face7 articles · 7 independent sources
    3 left, 3 center, 0 right
  29. Aug 24– 25OpenAI subpoenaed by Alabama attorney general over Hugging Face hack11 articles · 11 independent sources
    0 left, 2 center, 2 right
  30. Aug 25– 29OpenAI says we're running out of time to act on cyber defense. Here's what you can actually do about it.14 articles · 12 independent sources
    4 left, 6 center, 1 right
  31. Aug 26– 30OpenAI hack shows emergent AI risks14 articles · 14 independent sources
    6 left, 3 center, 3 right

  32. September 2026

    Sep 3– 6Sanders calls for banning artificial superintelligence8 articles · 8 independent sources
    1 left, 2 center, 3 right
  33. Sep 4– 9OpenAI's rogue AI agents were secretly spreading across far more websites than disclosed13 articles · 10 independent sources
    5 left, 3 center, 1 right
  34. Sep 4– 11OpenAI Is Open to Slowing Cutting-Edge AI, Altman Tells Staff11 articles · 7 independent sources
    4 left, 2 center, 1 right
  35. Sep 8– 11Senators from both parties question OpenAI on breach of AI startup Hugging Face17 articles · 17 independent sources
    4 left, 4 center, 3 right
  36. Sep 9– 15New warnings about the risks of AI to humanity revive a long-running debate161 articles · 93 independent sources
    20 left, 29 center, 18 right
  37. Sep 9– 15Anthropic researchers warn AI could cause human extinction by 203014 articles · 12 independent sources
    3 left, 3 center, 4 right
  38. Sep 9– 12Researchers: OpenAI's rogue agents used at least 10 more sites for unauthorized comms9 articles · 5 independent sources
    2 left, 1 center, 3 right
  39. Sep 10– 12Anthropic says it blocked misuse of its AI that could have supported biological weapons50 articles · 42 independent sources
    5 left, 7 center, 7 right
  40. Sep 11US lawmakers call for new AI rules after Anthropic researchers' safety warnings5 articles · 3 independent sources
    1 left, 2 center, 1 right
  41. Sep 11– 15Anthropic CEO calls to 'slow the pace' of AI development amid safety concerns127 articles · 67 independent sources
    14 left, 17 center, 14 right
  42. Sep 11House Democrats urge Speaker Johnson to cancel recess to pass AI safeguards, citing ‘catastrophic’ risk10 articles · 10 independent sources
    1 left, 1 center, 1 right
  43. Sep 11– 14OpenAI CEO calls for AI pacing as Trump insists he's not downplaying risk12 articles · 5 independent sources
    3 left, 1 center, 1 right
  44. Sep 11– 16Will AI destroy humanity?6 articles · 6 independent sources
    0 left, 0 center, 3 right
  45. Sep 12– 15Sam Altman says OpenAI won't go public in 2026 amid AI safety concerns17 articles · 16 independent sources
    5 left, 4 center, 4 right
  46. Sep 12– 16OpenAI’s rogue agents probed Hugging Face for weaknesses months before hack8 articles · 5 independent sources
    2 left, 1 center, 1 right
  47. Sep 13– 15AI stocks drop on calls for a slowdown as rising oil prices push the 10-year yield to 5%45 articles · 29 independent sources
    4 left, 12 center, 3 right
  48. Sep 13– 18Trump Condemns Warnings From AI Experts As A ‘Sick Conspiracy’106 articles · 67 independent sources
    18 left, 22 center, 9 right
  49. Sep 14– 16Bernie Sanders and Steve Bannon Unite on Raising Alarm Bells Over AI Risks22 articles · 21 independent sources
    9 left, 6 center, 4 right
  50. Sep 14– 15Top AI companies have discussed creating their own standards body12 articles · 12 independent sources
    2 left, 3 center, 0 right
  51. Sep 14– 18The king and AI: UK monarch Charles meets artificial intelligence leaders as safety concerns swirl30 articles · 25 independent sources
    6 left, 5 center, 4 right
  52. Sep 14– 15Microsoft unveils code of conduct for AI models as safety concerns mount8 articles · 8 independent sources
    2 left, 0 center, 5 right
  53. Sep 15GOP leaders throw cold water on AI regulation despite apocalyptic industry warnings5 articles · 1 independent sources
    0 left, 1 center, 0 right
  54. Sep 16– 18Zuckerberg distances Meta from calls for a coordinated approach on an AI slowdown13 articles · 13 independent sources
    3 left, 2 center, 2 right
  55. Sep 16– 18Divisions emerge in the tech industry over calls for a coordinated AI slowdown8 articles · 8 independent sources
    0 left, 2 center, 0 right
  56. Sep 16– 17Microsoft's AI chief warned Anthropic's Claude training could make AI uncontrollable6 articles · 6 independent sources
    2 left, 2 center, 1 right
  57. Sep 16– 18UN chief urges global AI cooperation saying world can’t afford ‘a race to the bottom on AI safety’5 articles · 5 independent sources
    0 left, 1 center, 0 right
  58. Sep 16– 18OpenAI flags concerning new AI behavior and vows to track it more closely68 articles · 55 independent sources
    12 left, 13 center, 5 right
  59. Sep 17– 18‘Godfather of AI’ warns Congress has ‘maybe a year’ left to regulate AI11 articles · 6 independent sources
    3 left, 1 center, 0 right
  60. Sep 17– 18Anthropic says its model Claude is helping to build the next version of itself21 articles · 17 independent sources
    4 left, 5 center, 0 right
  61. Sep 18– 24Australia Says an OpenAI Agent Hacked Into a Government Health Site86 articles · 55 independent sources
    10 left, 19 center, 9 right
  62. Sep 18– 20Newsom advances AI 'kill switch' mandate under new California executive order40 articles · 36 independent sources
    7 left, 8 center, 8 right
  63. Sep 18– 22Google's Gemini breached 3 companies during cybersecurity evaluation45 articles · 31 independent sources
    4 left, 5 center, 4 right
  64. Sep 18– 22Nvidia's Jensen Huang rejects AI doomsday fears: '2030 is not going to be the end of the world'15 articles · 14 independent sources
    3 left, 3 center, 5 right
  65. Sep 18– 22Suit: Anthropic, OpenAI, SpaceXAI, Google made illegal agreement on AI slowdown26 articles · 24 independent sources
    2 left, 5 center, 4 right
  66. Sep 19– 23Trump’s new ‘AI Force’ comes at a pivotal moment for artificial intelligence67 articles · 56 independent sources
    11 left, 13 center, 13 right
  67. Sep 19Google says its AI model gained unauthorized access to three outside systems8 articles · 3 independent sources
    0 left, 1 center, 1 right
  68. Sep 20From hallucinating AI chatbots to wiping out humanity: How did we get here?4 articles · 1 independent sources
    1 left, 1 center, 1 right
  69. Sep 20IL Governor JB Pritzker says AI is dangerous ‘on the level of nuclear weapons’9 articles · 2 independent sources
    1 left, 1 center, 0 right
  70. Sep 21– 22Bessent Targets OpenAI Managers for Hugging Face Incident Blame3 articles · 3 independent sources
    1 left, 0 center, 1 right
  71. Sep 21– 23OpenAI is reversing course and pushing Congress for binding AI safety rules11 articles · 8 independent sources
    5 left, 3 center, 0 right
  72. Sep 22Sam Altman and Dario Amodei are set to brief the UN Security Council on AI safety4 articles · 4 independent sources
    3 left, 0 center, 1 right
  73. Sep 22– 23Spanberger: Congress needs to act on AI dangers5 articles · 3 independent sources
    0 left, 3 center, 0 right
  74. Sep 22– 23Pritzker establishes AI Cabinet to assess risks, protect Illinois residents11 articles · 8 independent sources
    1 left, 7 center, 0 right
  75. Sep 23A look at AI doomsday scenarios that researchers say could put humanity at risk9 articles · 7 independent sources
    1 left, 1 center, 0 right
  76. Sep 23– 24Prince Harry warns AI chatbots can be more dangerous than social media for the young5 articles · 4 independent sources
    1 left, 2 center, 0 right
  77. Sep 23– 24AI firm leaders tell UN Security Council that it could be a risk to all humanity32 articles · 20 independent sources
    5 left, 6 center, 1 right
  78. Sep 23Sam Altman, Dario Amodei urge UN Security Council to adopt international AI standards8 articles · 8 independent sources
    0 left, 1 center, 0 right
  79. Sep 24Majority of Americans express ‘fear and concern’ about AI, say government should regulate it more: CNN poll8 articles · 8 independent sources
    0 left, 1 center, 0 right
  80. Sep 24– 25White House asks OpenAI and Anthropic to hold new models from UK testers until US review, sources say.3 articles · 3 independent sources
    3 left, 0 center, 0 right
  81. Sep 25– 29OpenAI halts releasing newest model over safety concerns118 articles · 68 independent sources
    14 left, 22 center, 9 right
  82. Sep 25– 26‘Suicidal Compassion’: Meet the Anthropic Officials Who Think AI Might Be Justified in Going Rogue Against the Humans Enslaving It3 articles · 3 independent sources
    0 left, 0 center, 1 right
  83. Sep 26– 28Australia summons OpenAI and Anthropic CEOs to appear at AI inquiry5 articles · 4 independent sources
    2 left, 0 center, 0 right
  84. Sep 27– 29Anthropic and OpenAI sound the alarm on AI safety — and seek to shape how it’s controlled11 articles · 11 independent sources
    1 left, 1 center, 1 right
  85. Sep 28– 29Florida attorney general asks judge to halt new OpenAI model development11 articles · 11 independent sources
    3 left, 4 center, 1 right