Multimodal AI Expansion and Privacy Controversies from Google and Meta

멀티모달 | Thu Jul 09 2026 00:00:00 GMT+0000 (Coordinated Universal Time) | 3 sources

SynthID watermark successfully detected a deepfake, Gemini powered a new video editing tool, and Meta AI glasses raised privacy concerns.

Analysis

[Google SynthID] successfully detected a McConnell deepfake image ^[1]

• Invisible watermark embedded in AI-generated images
• Snopes used SynthID to identify a manipulated image of McConnell on a sickbed
• Watermark persists through screencaptures and platform transfers
• Gemini participates from 2025
• OpenAI from May 2026
• while Anthropic does not participate

[Google Photos] launched Video Remix feature powered by Gemini Omni ^[2]

• AI video editing tool powered by the Gemini Omni model
• Cinematic relighting
• background replacement
• and watercolor
• oil painting
• and sketchbook style application
• Rolling out to Google AI Plus
• Pro
• and Ultra subscribers across 14 countries including the US
• South Korea
• Japan
• and India
• Editing possible with a few taps from the Create tab in Google Photos

[Meta AI Glasses] added camera deactivation feature when LED tampering is detected ^[3]

• Camera automatically deactivates if the recording indicator LED is tampered with
• Response to covert filming abuse cases involving taping over the LED
• Simultaneously testing prototypes with continuous audio recording and photo capture every few seconds
• Ongoing lawsuits over use of user images for Meta AI training and privacy violations

Sources

• [1] Google’s deepfake detector system used to debunk McConnell hoax pic - TechCrunch AI
• [2] Google Photos adds a new AI ‘Video Remix’ tool - TechCrunch AI
• [3] Meta wants its AI glasses to seem less creepy. Its AI strategy says otherwise. - TechCrunch AI