Italy's antitrust watchdog AGCM opened an investigation into Chinese AI startup DeepSeek for allegedly failing to give users "sufficiently clear, immediate and intelligible" warnings that its AI can produce "inaccurate, misleading or invented information." In February, Italy's data protection authority had ordered DeepSeek to block access to its chatbot after it failed to address privacy policy concerns.
A 40-second video of two students opposing Vice President Sara Duterte's impeachment, posted June 14 and viewed 7.3 million times, carried a watermark of Google's AI video generator Veo, Rappler found. Senator Bato dela Rosa, a Duterte ally, shared it; on June 16 Malacañang criticized him, saying disinformation from high-ranking officials further erodes trust.
A Houston Chronicle analysis found 11 of 13 cases directly quoted in a Houston Housing Authority brief, filed in a woman's lawsuit over her lost housing subsidy, did not contain the quotes. Kevin Fulton of Fulton Law Group cited a short deadline and did not say whether AI was used; the authority said it cannot confirm AI use but acknowledged quotations were not properly verified.
Indonesia's Ministry of Communication and Digital Affairs (Komdigi) kept its suspension of the World platform, saying its evaluation found personal data protection violations in Tools for Humanity's collection of iris biometrics via World ID. It ordered Tools for Humanity and local partner PT Sandina Abadi Nusantara to stop collecting and processing iris data, permanently delete Indonesians' iris codes, and ensure no children's data is processed.
Instagram users have complained for weeks of accounts wrongly banned or suspended, some for alleged child sexual exploitation, with some saying appeals went unanswered. Many suspect, without direct evidence, AI moderation; a Change.org petition has over 4,000 signatures, some users threaten a class action, and Meta has not publicly acknowledged the issue.
Several expert reports ordered by German courts and seen by Spiegel found Tesla's camera-based 'Tesla Vision' parking assist often fails, at times missing simple obstacles such as a box, a bike rack or a child sitting in front of the car. One expert judged it 'not equivalent' to ultrasonic sensors and another found a Peugeot 3008 more reliable; Tesla did not respond to Spiegel.
Opendoor agreed to pay $39 million to settle a 2022 investor lawsuit alleging it misrepresented its pricing algorithm as an AI-powered system that could outpace traditional real estate firms when, the suit claims, it relied heavily on human input. The proposed settlement needs approval from U.S. District Judge Michael Liburdi; Opendoor admitted no wrongdoing.
Parents who paid 8,000 yuan for Xueersi AI learning machines say a feature in which human teachers recorded video answers within an hour was replaced by 'AI Ask Anytime', which they say misreads questions and sometimes explains backwards from wrong answers. One parent says it went offline without notice when the free period ended, and complaints to customer service went unanswered.
