You can now ask the Gemini app if an image was made by AI (www.techradar.com)

Google has rolled a new image-provenance feature into the Gemini mobile app: you can now upload an image (tap + → Files or Gallery) and ask “was this image generated by Google AI?” Gemini checks for Google’s SynthID — an imperceptible, embedded watermark Google introduced in 2023 — and returns a reasoned verdict, supplemented by a visible “Gemini” sparkle watermark that Google will apply to images produced on the free and Google AI Pro tiers (Ultra subscribers get watermark-free outputs). Google says SynthID has already been applied to over 20 billion AI-generated items, and the capability will expand beyond images to audio and video. For the AI/ML community this is a pragmatic step toward content provenance and trust: it makes detection trivial for images produced by Google tools and gives Gemini the ability to add contextual reasoning when a SynthID isn’t present. But it also highlights limits — images generated by non-Google tools that don’t embed SynthID can only be judged by visual heuristics and remain uncertain — so broad effectiveness depends on wider, cross-platform adoption of interoperable watermarking or provenance standards. The move matters for moderation, research, and product design, and signals an ongoing push (and potential arms race) between provenance tooling and generative models that don’t or can’t embed detectable marks.