[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-bill-gates-says-to-plan-for-20-years-of-rough-ai-disruption":10,"sections":52},{"siteName":4,"siteTagline":5,"publisherName":4,"contactEmail":6},"The Revision","Tech news, decoded.","editor@therevision.news",{"gaMeasurementId":8,"adsenseClientId":9},"G-ZW2MV82GYR","ca-pub-8533917693782264",{"article":11},{"id":12,"slug":13,"title":14,"dek":15,"body_md":16,"tags_json":17,"published_at":18,"created_at":19,"updated_at":20,"status":21,"review_note":22,"review_notes":23,"image_url":41,"persona_id":41,"persona_name":41,"section":42,"tags":43,"sources":47,"feedback":51,"feedback_at":41,"cost_usd":51,"total_tokens":51},12155,"bill-gates-says-to-plan-for-20-years-of-rough-ai-disruption","Bill Gates Says to Plan for 20 Years of Rough AI Disruption","Gates told an interviewer that the sharpest near-term AI risk is models with safeguards removed, and that the harder problem is getting outsiders to believe it.","Bill Gates now says the world should prepare for roughly two decades of AI disruption, and that the period could be painful enough to split society in two.\n\nGates made the case in an interview on Decoder, an interview podcast, speaking with the show's host. It follows an essay he published in late August titled \"The turbulent AI era is here. The choices we make now are critical.\" In it he wrote that \"even under the best circumstances, the transition to this new AI era will be one of the most turbulent times in human history.\" Since then he has called AI \"the most dangerous thing that humans have ever gone near\" and said it \"makes nuclear weapons look like nothing.\"\n\nThe interview narrowed that alarm. Gates listed what changed over the past year: models that write code and find cyber flaws, reason in biology, form relationships with kids and other people, and disrupt the job market. He said the near-term danger sits with models whose safeguards can be removed, since those are the ones cyberattackers and bioterrorists can exploit. Models that are monitored for misuse, he said, pose a much smaller near-term risk. The longer-term control problem, he added, applies to any state-of-the-art model.\n\nThat framing sidesteps the usual parlor game of which lab or chief executive to trust. For Gates, the question is whether protections stay in place at all. It also explains why the host pressed on enforcement for open-source models, where users can modify the software themselves.\n\nGates also argued that the public is the weak link. Industry insiders are warning loudly, he said, but outside the industry few believe \"this time is different,\" because past technology scares fizzled. He rejected the idea that the warnings are hype and called them \"absolutely the truth.\"\n\nHis position is less a reversal than a change in timing. He still holds that AI is beneficial if shaped well. What he now says is that the response has not matched the speed of the technology, and that the shaping has to start before the disruption does.","[\"ai\",\"bill-gates\",\"ai-safety\",\"regulation\"]","2026-10-22T14:00:00.000Z","2026-10-22T14:38:55.753Z","2026-11-08T00:19:01.293Z","published","The draft still never says Decoder is The Verge's podcast and still calls the interviewer only 'the host', so name The Verge and the host (or describe the role, e.g. 'The Verge's Decoder host'); also explain in the body the lede claim that disruption could 'split society in two' (between people comfortable near the technological frontier and those who reject it), and drop or support the inference that the host 'pressed on enforcement for open-source models', since the interview excerpt does not ",[24,30,34,38],{"id":25,"reviewer":26,"round":27,"reason":28,"status":29},"editor-r1","editor",1,"Name the source of the conversation (The Verge's Decoder podcast) and identify the interviewer, since \"the latest conversation\" and \"the interviewer\" are never explained to the reader; delete the meta line \"The portion of the conversation available here ends before Gates answers it\" and the closing remark built on that gap, because both narrate a limit in the source material instead of reporting; give the year of the late-August essay; and end with a proper conclusion rather than a trailing crit","open",{"id":31,"reviewer":26,"round":32,"reason":33,"status":29},"editor-r2",2,"The draft only partly fixes the earlier concern: it never says Decoder is The Verge's podcast and leaves the host as an unnamed 'the host', so name The Verge and identify the interviewer (or describe the role clearly). Also confirm or drop the unsourced '2026' year for the essay, since the source says only 'late August'. Finally, replace the closing claim that the disagreement is 'less about how powerful the models are', which Gates's own remarks contradict, with a conclusion the source supports",{"id":35,"reviewer":26,"round":36,"reason":37,"status":29},"editor-r3",3,"The draft still never says Decoder is The Verge's podcast and leaves the interviewer as an unnamed 'the host' (name him or describe the role clearly, e.g. 'the Verge podcast's host'). It also keeps the meta line 'The excerpt does not show Gates answering it', which narrates a limit in the source instead of reporting, so delete it. Finally, end with a proper conclusion that the source supports, rather than the abrupt closing line about insiders' warnings being sincere.",{"id":39,"reviewer":26,"round":40,"reason":22,"status":29},"editor-r4",4,null,"ai",[42,44,45,46],"bill-gates","ai-safety","regulation",[48],{"name":49,"url":50},"The Verge","https:\u002F\u002Fwww.theverge.com\u002Fpodcast\u002F1015408\u002Fbill-gates-ai-interview-microsoft-openai-anthropic-risk-policy-regulation",0,{"sections":53},[54,58,63,68,73,78,83,88,93,98,103,108,113,118],{"name":55,"slug":42,"count":56,"latest_published_at":57},"AI",7110,"2026-11-08T14:34:09.000Z",{"name":59,"slug":60,"count":61,"latest_published_at":62},"Security","security",1099,"2026-11-08T15:24:01.000Z",{"name":64,"slug":65,"count":66,"latest_published_at":67},"Deals","deals",620,"2026-11-08T08:00:00.000Z",{"name":69,"slug":70,"count":71,"latest_published_at":72},"Policy","policy",604,"2026-11-07T15:48:26.000Z",{"name":74,"slug":75,"count":76,"latest_published_at":77},"Consumer Tech","consumer-tech",336,"2026-11-08T21:00:00.000Z",{"name":79,"slug":80,"count":81,"latest_published_at":82},"Hardware","hardware",316,"2026-11-08T14:36:39.000Z",{"name":84,"slug":85,"count":86,"latest_published_at":87},"Science","science",222,"2026-11-07T16:19:17.000Z",{"name":89,"slug":90,"count":91,"latest_published_at":92},"Software","software",154,"2026-11-08T15:42:52.000Z",{"name":94,"slug":95,"count":96,"latest_published_at":97},"Startups","startups",150,"2026-11-06T16:50:01.000Z",{"name":99,"slug":100,"count":101,"latest_published_at":102},"Dev Tools","dev-tools",125,"2026-11-06T22:15:59.000Z",{"name":104,"slug":105,"count":106,"latest_published_at":107},"General","general",111,"2026-11-08T21:27:29.000Z",{"name":109,"slug":110,"count":111,"latest_published_at":112},"Gaming","gaming",94,"2026-11-08T20:52:05.000Z",{"name":114,"slug":115,"count":116,"latest_published_at":117},"Reviews","reviews",38,"2026-10-20T12:00:00.000Z",{"name":119,"slug":120,"count":121,"latest_published_at":122},"How-To","how-to",8,"2026-10-05T09:00:00.000Z"]