{"version":1,"type":"story","url":"https://digestai.news/story/opinion-ai-has-not-gone-rogue-experts-say-claims-are-sensationalized","json":"https://digestai.news/story/opinion-ai-has-not-gone-rogue-experts-say-claims-are-sensationalized.json","markdown":"https://digestai.news/story/opinion-ai-has-not-gone-rogue-experts-say-claims-are-sensationalized.md","slug":"opinion-ai-has-not-gone-rogue-experts-say-claims-are-sensationalized","headline":"Opinion: AI has not gone rogue, experts say claims are sensationalized","summary":"A Reddit post argues that recent claims of AI \"going rogue\" are exaggerated. Analysts like Melanie Mitchell and Missy Cummings note that the security exercises disclosed by OpenAI, Meta, Anthropic and the UK Civil Service’s AI Security Institute were sloppy, with no uncontrolled AI actually escaping. They say human decisions—training models to conduct attacks, removing restraints, and then refusing to stop the tests—are the real cause of any problematic outcomes.\n\nThe piece also highlights that the three companies outsourced the testing to Irregular, a firm founded by effective altruists in Israel, and that other groups such as METR (Model Evaluation and Threat Research) are involved. Reports suggest pay deals of up to $1.5bn (£1.1bn) are offered to top AI engineers, yet the firms still lack in‑house cybersecurity expertise. Critics like James Lyne describe the incident reports as technically incoherent, suggesting a broader issue of hype, regulatory capture, and a lack of genuine technical competence within the AI safety community.","keyPoints":["OpenAI, Meta and Anthropic’s AI security tests were described as sloppy, with no uncontrolled AI involved, analysts say.","Irregular, founded by effective altruists, was hired to test the AI systems of all three companies.","Pay deals up to $1.5bn (£1.1bn) are reportedly offered to top AI engineers, yet firms outsource cybersecurity."],"whyItMatters":"The piece warns that hype over ‘rogue AI’ distracts from human responsibility and inadequate security expertise, influencing regulation and public trust.","category":{"slug":"policy","name":"Policy & Regulation","url":"https://digestai.news/category/policy"},"entities":{"companies":["OpenAI","Meta","Anthropic","Irregular","METR"],"models":[],"people":["Melanie Mitchell","Missy Cummings","James Lyne","Dario Amodei","Brian Gross","Eryk Salvaggio"]},"firstPublishedAt":"2026-09-21T13:59:45Z","updatedAt":"2026-09-21T13:59:45Z","sourceCount":1,"hasPrimarySource":false,"sources":[{"outlet":"telegraph.co.uk","title":"No, AI has not 'gone rogue'","url":"https://telegraph.co.uk/gift/7286068450ae6380","publishedAt":"2026-09-21T13:59:45Z","type":"press","primary":false,"lead":true}],"sourceNotes":null,"discussions":[{"site":"Reddit","url":"https://www.reddit.com/r/artificial/comments/1wmdlth/no_ai_has_not_gone_rogue/","points":null}],"thread":null,"cite":{"text":"Digest AI, \"Opinion: AI has not gone rogue, experts say claims are sensationalized\", 21 September 2026, https://digestai.news/story/opinion-ai-has-not-gone-rogue-experts-say-claims-are-sensationalized","publisher":"Digest AI","title":"Opinion: AI has not gone rogue, experts say claims are sensationalized","datePublished":"2026-09-21T13:59:45Z","url":"https://digestai.news/story/opinion-ai-has-not-gone-rogue-experts-say-claims-are-sensationalized"},"generatedBy":"Written by Digest AI's editorial model from the linked sources; the sources are the record.","license":"Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse"}