Tesla gave Swedish and Dutch regulators self-compiled safety statistics that independent road-safety researchers told Reuters amount to misleading marketing. An April slide deck sent to Swedish regulators claimed Full Self-Driving cars travel over seven times farther between crashes than average US drivers and could have saved 32,000 lives, figures researchers say assume every US vehicle, trucks included, becomes a Tesla.
South Korean police said a six-month cyber sexual violence crackdown from November 2025 produced 1,506 arrests, 87 under detention, with teenagers accounting for 723, or 42.9%. The sweep targeted exploitation material and illegal footage broadly; one suspect accused of making deepfakes from students' photos and running phishing scams impersonating investigators was caught and detained in Malaysia.
A Consumer Reports investigation found Uber and Lyft quoted markedly different fares to riders requesting the same route at the same minute, including 29 prices for 55 would-be passengers on one Kansas City trip, and that about 11% of apparent discounts rested on seemingly inflated original prices. The companies challenged the findings; Uber calls the struck-through figures "historical comparison messaging."
The Institute for the Study of War says it has grounds to believe that videos published on 15 June of Russian troops raising flags in and around Kostyantynivka may be AI-generated, in line with earlier Russian information operations. Russia's defence ministry claimed nearby Dovha Balka the same day, while a Ukrainian corps commander said Russia is running information operations about the city.
White House officials told The Washington Post that Anthropic belatedly disclosed about 50 organisations with early Mythos access beyond 111 the administration had approved, one a South Korean telecom suspected of China ties; Anthropic reportedly revoked that access quickly. The US has since issued export-control guidance barring foreign access to Mythos 5 and Fable 5, and Anthropic suspended both for all users.
Chief Judge Martin Reidlinger (W.D.N.C.) declined to sanction plaintiff's counsel Fred W. DeVore III and Rob Wilder, who admitted that briefs citing two nonexistent cases and invented quotations stemmed largely from AI misuse and unverified output. Their agreed mea culpa in the N.C. State Bar Journal fell short and minimised the errors, the judge wrote, but their repentance was made in good faith.
In a Brooklyn construction-fall suit, Justice Devin P. Cohen of the Kings County Supreme Court found that of about 38 cases in defense counsel's papers, 14 either do not exist or are not at the citation given, some with names resembling real cases. A Westlaw editor's note says the citations were likely AI-generated.
An Instagram account run by a sports page posted an AI-edited image giving a World Cup presenter a greatly enlarged bust. After an influencer flagged it, Sportschau's official account replied that it was striking such content still finds favour in 2026, and other users, including another presenter, condemned the post.
A Greek television journalist, back on the show "Happy Day" after her husband's medical ordeal, said AI-generated images had circulated online showing her crying inside and outside intensive care and her husband in a wheelchair and intubated. She told viewers the children of public figures should hear the facts at home rather than believe what they see.
More than a hundred people and organisations, including Amnesty International, Access Now and the Stop Killer Robots campaign, signed a statement warning that AI in military "kill chains" risks civilian bloodshed and unaccountability. It cites Israel's use of Lavender, Gospel and Where's Daddy in Gaza, which it says may obscure international crimes behind "perceived algorithmic objectivity", and urges tech firms to stop supplying militaries.
Microsoft researchers and other security firms report attackers using EvilTokens, a phishing-as-a-service platform combining AI and automation, to trick staff into entering attacker-generated device codes on Microsoft's genuine login page, bypassing multi-factor authentication to take over accounts. Taiwan's National Institute of Cyber Security issued a warning; Huntress says Microsoft environments face greater risk than Google's.
