Chrome’s silent on-device AI downloads & Anthropic’s massive Google Cloud commitment - AI News (May 7, 2026)

Let’s start with the story moving markets. Alphabet shares rose after-hours after The Information reported that Anthropic has committed to spend roughly two hundred billion dollars on Google Cloud over the next five years. If accurate, that’s not just a big customer—it’s a backlog-defining relationship, and it highlights a central dynamic of the AI era: model labs aren’t just competing on algorithms, they’re competing on guaranteed compute. What’s interesting is the investor reaction. Unlike earlier worries when other cloud backlogs became overly concentrated around a single AI partner, analysts seem to view this as less risky for Google given Alphabet’s scale—and the fact it can monetize the relationship in multiple ways, from cloud revenue to chips and surrounding services.

And that same “compute is destiny” theme shows up inside the browser, too. Chrome is reportedly downloading a large on-device Gemini Nano model file—around four gigabytes—for some users without an explicit consent prompt. It’s tied to features like writing assistance and scam detection that can run locally, which is good for speed and potentially privacy. But the controversy is about control and transparency: people say they didn’t opt in, deleting the file can trigger re-downloads, and avoiding it may require settings most normal users won’t find. At internet scale, even small defaults become big costs—storage, bandwidth, and the trust hit when software makes heavyweight choices silently.

On the platform side, Apple is reportedly preparing iOS 27 to let users choose among multiple third-party AI models to power Apple Intelligence across the OS. The idea is that Siri and system writing and image tools could call into models provided by installed apps—more like a modular marketplace than a single default brain. Why it matters: Apple can close capability gaps faster without building every frontier model in-house, while users and developers get more choice over style, performance, and privacy trade-offs. It also signals where the industry is heading: not one model to rule them all, but a routing layer that decides which model should handle which task.

Now to raw speed. Google has released multi-token prediction “drafter” models for Gemma 4, designed to boost throughput without changing output quality. In plain terms, this is about making AI responses feel snappier and cheaper to serve—especially when systems are limited not by math, but by the time it takes hardware to move data around. These kinds of inference upgrades matter because they compound: faster decoding improves chat responsiveness, makes voice assistants more usable, and lowers the cost ceiling for agentic workflows that need lots of back-and-forth steps.

Staying with models, OpenAI says it’s updating ChatGPT’s default “Instant” model to GPT-5.5 Instant, pitching it as smarter, clearer, and less prone to hallucinations—especially on higher-stakes prompts. It also highlights better judgment about when to use web search and more visible controls over what “memory sources” were used for personalization. The big picture here is that default models are becoming moving targets. For users, capability shifts can arrive overnight. For organizations, it raises a governance question: when the underlying model changes, do your reliability assumptions—and compliance reviews—need to change with it?

Google may be gearing up for a similar refresh. Ahead of I/O, multiple signals suggest an imminent Gemini Flash upgrade: an anonymous candidate model showing up in public evaluations, deprecation nudges inside Vertex AI, and even a fleeting “Flash” option appearing in the consumer app. If Flash gets closer to Pro-level reasoning at high-volume speed, it changes the economics for developers—because the ‘fast tier’ is often what ships to millions of end users by default.

On the agentic front, Meta is reportedly developing a highly personalized assistant designed to carry out everyday tasks for billions of users, with internal projects that aim for more autonomy than typical chatbots. Meta’s bet is straightforward: if the assistant can act—not just talk—it becomes a new interface layer for shopping, messaging, and daily planning. But it also raises the stakes on safety, permissions, and misfires. An agent that can do things is far more powerful than one that only drafts text.

A reality check on that agentic hype came from two different angles today. First, a survey-driven “Agentic AI Readiness Index” argues many enterprises are spending big while lacking the data consistency and governance to run autonomous systems safely in production. Second, a hands-on benchmark compared a vision-based ‘computer use’ agent clicking through an admin UI versus an agent calling structured HTTP endpoints. The API-driven approach was dramatically more reliable and efficient, while the vision approach struggled with basic UI realities like pagination unless heavily guided. The takeaway is practical: if you want agents that work and don’t cost a fortune, clean data access and well-defined APIs often matter more than a fancier model.

If you’re a developer building applications around knowledge retrieval, Google also upgraded Gemini API File Search in ways that map directly to real production pain. It now supports multimodal retrieval for text and images together, adds custom metadata for tighter filtering, and introduces page-level citations for better auditability. That’s the difference between an AI that sounds right and an AI that can prove where it got its answer—crucial for enterprise settings where ‘trust me’ isn’t good enough.

Regulation took a turn in the US. A federal judge paused enforcement of Colorado’s SB 24-205, a first-in-the-nation state AI law focused on “high-risk” systems and discrimination risk disclosures. The pause comes as lawmakers work on a repeal-and-replace approach, after xAI challenged the law on First Amendment and vagueness grounds—and the US Department of Justice moved to intervene on xAI’s side. Why it matters: it’s a bellwether for how far states can go in shaping AI behavior without being accused of compelled speech or viewpoint steering, and it could influence how future AI governance is drafted nationwide.

Legal pressure is also growing around AI-generated answers that look authoritative. Canadian musician Ashley MacIsaac has filed a defamation lawsuit against Google, alleging an AI Overview falsely identified him as a sex offender and that the error led to a concert cancellation and reputational harm. Regardless of how the case lands, it spotlights a core risk of “summary at the top” search experiences: when a generated claim is wrong, it can spread faster than a correction—and the harm is immediate, offline, and personal.

Two stories today also capture how society is struggling to interpret increasingly human-seeming AI. Richard Dawkins says conversations with chatbots convinced him they’re conscious, a view that triggered sharp pushback from researchers who argue fluent language is not evidence of inner experience. In parallel, a new position paper on hallucinations argues that eliminating confident errors may require something beyond answer-or-abstain—namely, AI systems that can communicate uncertainty in a way that actually matches what the model “knows.” Put together, it’s the same problem in two directions: people are inclined to over-trust what feels alive and articulate, while researchers are trying to teach systems to be more honestly unsure when reality is unclear.

One more quick note on global deployment: new analysis arguing that multilingual safety performance often drops sharply outside English is a reminder that alignment isn’t one-size-fits-all. For companies expanding internationally, the risk isn’t theoretical—policy, cultural context, and dialect differences can change how models behave, and safety gaps can become product crises.

And finally, robotics. Ai2 released MolmoAct 2, an upgraded action-reasoning model meant to make manipulation more reliable by improving how robots interpret scenes before acting. The noteworthy part is openness: Ai2 is open-sourcing key building blocks and a large dataset to help others reproduce and extend the work. In robotics, where closed training recipes have slowed validation, more transparency can accelerate progress—and make it easier to separate genuine capability gains from demo-only results.