NOTE: WDGDTW is moving to Beehiiv soon (sorry for the delay, tons of travel and conferences getting in the way!). Subscribe now before the price increases.
No full issue this week due to a migraine but Google continues to fine-tune Gemini across its product set while treading carefully around growing friction in the AI ecosystem. I summed it up here. The latest Gemini 2.5 Pro Preview includes long-requested productivity upgrades, most notably, support for scheduled actions and planned tasks across devices. It's a shift toward agent-like behaviour, as Google inches Gemini closer to becoming a proactive assistant, not just a reactive chatbot.
But product ambition is colliding with ecosystem fallout. Google has paused the rollout of Ask Photos, a Gemini-powered AI search for Google Photos—after early testers raised issues around accuracy and context handling. It’s a reality check that not all AI features scale cleanly across Google’s legacy platforms.
Meanwhile, Google’s AI partner Anthropic is in hot water, facing a lawsuit from Reddit over training on its content without a license. The case throws a spotlight back on Google’s own data practices and the broader fragility of the LLM supply chain. And in the background, DeepMind CEO Demis Hassabis is publicly reframing AI as a force for moral improvement, an aspirational narrative that’s increasingly out of sync with legal and platform tensions brewing around the company’s real-world deployments. While he had a full crowd at SXSW London little of substance was given.
Stay ahead with the full 'What Did ____ Do This Week?' ecosystem. Choose from OpenAI, Amazon now, and Meta soon.