In the spirit of keeping tabs on AI, fake account and disinformation detector Cyabra is now on the track of deepfakes — convincing, AI-powered media manipulations. Its two proprietary AI models, PixelProof for images and MotionProof for video, aim to detect tiny pixel, motion and lip-sync inconsistencies. Cyabra expects the solutions to be of interest to governments as well as brands.
Mobile marketing analytics platform AppsFlyer has adopted a model context protocol, the solution developed by Anthropic that helps AI communicate with external tools and systems. The MCP allows a user’s LLM of choice (ChatGPT, Claude, etc) to query AppsFlyer data, including attribution, analytics and audiences. The aim is to make those data insights available through natural language conversations.
Has AI entered the “trough of disillusionment”? Apparently not: of 600 B2C marketers surveyed by Invoca, 94% report AI-powered revenue growth and 80% were delighted with ROI (only 3% were unhappy with it). They’re pretty pleased with themselves as well: 92% say they’re AI experts or advanced users.
CDP Zeta Global has unveiled Athena, a “superintelligent” voice-activated interface that delivers instant answers, smarter decisions, and agentic actions throughout the Zeta platform. It activates a suite of agentic apps that can adapt to each user’s specific goals and styles. Early access will begin later this year, with a suite of agentic apps to follow in the first half of next year.