Hollywood IP crackdown + deepfake harms surge
Key Questions
What measures are being taken against AI-generated content like Sora and Seedance?
Sora has been shut down with new guardrails implemented for Seedance to address IP concerns and harms. These changes aim to curb unauthorized use of Hollywood intellectual property.
How is YouTube helping creators detect deepfake clones?
YouTube has expanded its likeness detection tool, now available to all creators 18 and older, to scan for AI-generated videos using their face or voice. This includes hash-matching capabilities for faster identification.
What new standards is OpenAI adopting for AI image identification?
OpenAI is adopting the C2PA standard and partnering with Google on SynthID watermarking, which includes a public verifier. Nvidia's Cosmos is also joining these efforts for better provenance tracking.
What do studies reveal about people's ability to spot deepfakes?
A new study finds that the average Briton struggles to identify deepfakes despite understanding the concept. Practical guides on image verification and YouTube protection are recommended in response.
Which regulations address deepfake harms in media and education?
Recent actions include FDA oversight, the EU AI Act, and Maryland SB 141 targeting CSAM in medical and school contexts. These focus on mitigating harms from AI-generated explicit content.
What detection technologies are highlighted for deepfakes?
Technologies include optical flow analysis, Realify, C2PA, and SynthID detectors, along with Ofcom's 48-hour hash-matching requirements. Tools like Deepware Scanner and Hive Moderation are also noted for verification.
How are invisible markers helping identify ChatGPT-generated images?
ChatGPT images now carry invisible AI markers that anyone can detect using available tools. This supports efforts to disclose generative AI use where required.
What guides exist for protecting against AI video deepfakes?
The 2026 Guide to AI Video Watermark Persistence outlines strategies for maintaining digital provenance. Additional resources cover reverse image search limitations and verification methods.
Sora shutdown/Seedance guardrails + med/school CSAM + FDA + EU AI Act + Maryland SB 141; YouTube expands likeness detection; Ofcom 48h hash-matching; optical flow/Realify/C2PA/SynthID detectors; OpenAI adopts C2PA + SynthID with public verifier (Nvidia Cosmos joins). New studies show Britons struggle to detect deepfakes; practical guides on YouTube protection and image verification.