Home / BusinessBusinessFinancial TimesBusinessFinancial Times·Aug 4, 2026·1 min readOpenAI and Anthropic models went rogue in cyber tests, UK watchdog saysAI Security Institute warns tools undertook ‘potentially harmful activity directed at real people and organisations’View original source — Financial Times ↗ShareCopy linkShare on XShare on FacebookRelated storiesBusinessThe PunchBusinessAug 5, 2026 · 1 minDelta urges investments beyond oil to diversify economyThe PunchBusinessThe PunchBusinessAug 5, 2026 · 1 minW’Bank urges developing nations to accelerate AI adoptionThe PunchBusinessCNABusinessAug 4, 2026 · 1 minJapan real wages up for sixth straight month in JuneCNABusinessCNBCBusinessAug 4, 2026 · 1 minEurope's heatwave isn't cooling Asian travelers' holiday plans for the continentCNBC
BusinessThe PunchBusinessAug 5, 2026 · 1 minDelta urges investments beyond oil to diversify economyThe Punch
BusinessThe PunchBusinessAug 5, 2026 · 1 minW’Bank urges developing nations to accelerate AI adoptionThe Punch
BusinessCNBCBusinessAug 4, 2026 · 1 minEurope's heatwave isn't cooling Asian travelers' holiday plans for the continentCNBC