1. Microsoft's planning docs for Scout say the goal is to "make people addicted." DuckDuckGo just shipped a button to turn AI off. Internal planning documents for Scout, Microsoft's new AI assistant, state that the plan is to "make people addicted" to the tool before adding new features, according to 404 Media, which reviewed

2. The AI vendors said "use it everywhere." Uber's budget lasted four months. Uber told its staff to use AI as much as possible. Four months later, the company is capping how much they can spend, according to TechCrunch, after employees blew through the budget set aside for it.

3. Anthropic Sent Its Vulnerability-Hunting Model Into Power Grids and Hospitals in 15 Countries The model Anthropic built to find security holes now sits inside the systems where a single breach can reach 100 million people.

In Brief

• Trump signs executive order for pre-release review of frontier AI models President Trump created a voluntary framework letting AI companies share frontier models with the federal government before release. The order frames the review as protecting critical infrastructure cybersecurity while avoiding mandatory rules. Participation stays optional.

• OpenAI ships six job-specific Codex plug-ins OpenAI released Codex plug-ins targeting data analytics, creative production, sales, product design, equity investing, and investment banking. Each bundles integrations, instructions, and context so Codex approximates a specific white-collar role. The tools run inside the Codex app.

• Microsoft launches MAI-Thinking-1, its first reasoning model Microsoft introduced MAI-Thinking-1 as its flagship in-house reasoning model at Build 2026. The company began building its own models last year after relying on OpenAI, and the two recently renegotiated their deal to loosen ties.

• Microsoft unveils Project Solara, an Android-based OS for agent gadgets Microsoft announced Project Solara at Build, a platform for devices that run AI agents. It is built on Android rather than Windows. Microsoft showed two concept devices: a desk unit and a wearable badge.

• Google adds fake call detection against AI voice scams Google rolled out detection for calls that spoof trusted numbers and use AI deepfakes to imitate authority figures, family members, or employers. The feature responds to scammers shifting tactics as people stop answering unknown numbers.

• Microsoft gives developers portable policy files for agent behavior Microsoft released a specification that lets developer, compliance, and security teams define rules for AI agents in portable policy files. The files travel with agents to enforce constraints across deployments.

• Google's Gemini Spark plans trips as an autonomous agent Google's Gemini Spark searches travel options, researches activities, and assembles itineraries from a single prompt. The Verge tested it on trip planning, the use case AI vendors have promised for four years.

• Cyera nears $300M round at $12B valuation Cybersecurity firm Cyera is closing a round led by Evolution Equity Partners that values it at $12 billion, an 80x ARR multiple. The company posts operating losses despite the valuation.

• Amazon faces class action over Ring facial recognition A Virginia resident sued Amazon in Seattle over Ring's Familiar Faces feature. The complaint claims Ring stores images of passersby without their consent.

• The Economist questions whether public markets can absorb AI's giants The Economist examined whether stock markets can handle Anthropic, SpaceX, and OpenAI entering public trading. The piece weighs the combined valuations against available investor capital.

• Martin Scorsese uses AI for storyboarding Director Martin Scorsese endorsed AI, becoming an unexpected Hollywood voice for the technology. He limits his use to storyboarding rather than production or final footage.

Read the full edition →