OpenAI and rivals show image generators producing cartoon or realistic election imagery; tests reveal varying adherence to safeguards, with cartoons reducing evidentiary value but realistic fakes posing deception risks; audits flag potential misuse and need for watermarking or safeguards.
Translated title. Original: Videos virales con IA: cómo poner tu cara en TikTok
AI face-swap videos let users insert their face into viral scenes for short-form platforms; the article explains how to create such videos, cautions about potential deception, and advises on privacy, consent, and labeling AI-generated content.
Microsoft Teams will add real-time, third-party deepfake detection in November 2026, allowing certified security vendors to scan live video and audio, provide granular risk scores, and display warnings within meeting windows to combat impersonation and social engineering. The rollout includes enterprise-focused integration with SIEM telemetry and an accompanying in-house impersonation protection feature.
Translated title. Original: Рассвет дипфейков: как ИИ изменил порно, политику и реальность - BBC News Русская служба
The article reports on the spread and financial costs of deepfakes, which can imitate real people in images, video, and audio. It describes how they are making it harder to tell authentic content from fakes in everyday life and politics.
Translated title. Original: La guerra de los famosos contra la IA: 3.000 euros al mes para borrar el contenido falso | Mujerhoy
AI-generated fake content about celebrities is growing, prompting a market for deletion and reputation protection services that charge thousands per month; the industry uses verification and removal tactics, while fakes can also influence politics and media narratives.
Translated title. Original: DeepSFU repère les deepfakes sans décoder chaque image
DeepSFU detects deepfakes by analyzing compressed video data rather than full-resolution frames, achieving near-perfect accuracy on 1,000 videos and requiring far less computation. The method aims to enable real-time detection in large-scale videoconferencing, with plans to adapt for widespread use despite current limitations in real-time defense within common conferencing tools.
Translated title. Original: Österreichische Verbände gegen pauschale KI-Kennzeichnung - epd medien
Thirteen Austrian industry associations say AI-generated advertising should not require a label simply because it uses AI. They argue labeling should focus on ads that could mislead audiences into mistaking a deepfake for reality.
Translated title. Original: Fejkade AI-affischer hetsar mot förorten - Nyhetsbyrån Järva
Fake AI-generated posters spread online to incite protests in Järva, mixing false origins and violent rhetoric; fact-checking traces origins to unclear sources and possible right-wing actors, while real anti-fascist mobilization days are also being discussed but no valid demonstrations are confirmed.
Translated title. Original: Intelligence artificielle : non, un lion n'a pas sauvé un enfant d'une hyène - ICI
A doctored AI-generated video claiming a lion saves a child from a hyena circulated on social media, drawing mixed reactions and prompting debunking efforts. Such sensational clips aim to generate clicks and revenue, with prior examples like a gorilla rescue; organizations have traced origins to AI content creators and advise verification.
Translated title. Original: SVT lurade SVT – djurklipp var AI
SVT tested its local newsrooms with fabricated tips and AI-generated animal videos; three newsrooms published the clips before learning they were fake. Editors said the test exposed failures in verification, though SVT could not confirm whether the videos spread further.