POLICY

Will EU's Google ruling make AI better for all?

The EU is forcing Google to play nice with its AI competitors. 

On Thursday, the European Commission ordered the search giant to share data with other search engines and to open up the Android operating system to rival AI companies, taking sweeping measures to curb Google's dominance under the Digital Markets Act. The EU said the ruling is an attempt to increase competition and diversify AI innovation, as well as give users more choice. 

The first part of the decision opens 11 features on the Android OS to AI rivals, allowing them to better compete with Gemini. 

• For instance, users will now be able to activate competing AI assistants, such as those from OpenAI or Anthropic, via voice commands, similar to the "Hey Google" feature. 

• These features, however, will only be available to rivals that meet security and privacy requirements protecting users' data and device safety. The change will go into effect in July 2027 as part of the next iteration of Android. 

• In its announcement, the European Commission said that the closed Android OS limits how useful competing AI services can be, making it difficult for the 60% of EU users who have Android devices to actually benefit from them. 

The second aspect of the ruling compels Google to share data with competing search services, aiming to create a "more level playing field with Google Search." This means that AI chatbots offering search functionality are able to receive shared data. This data is subject to anonymization and allows Google to assess whether sharing certain data poses cybersecurity and data protection risks.

Teresa Ribera, executive VP for clean, just and competitive transition at the commission, said in a statement that the decision will help smaller firms and AI services compete with tech giants and provide greater choice to users. "Society is going through a profound digital transformation," said Ribera. "We need to keep that process fair and ensure that our citizens have choice." 

Google, however, criticized the decisions, claiming that it risks undermining "vital privacy and ⁠security guardrails" for millions of European citizens, the company's lawyer Kent Walker told Reuters. "These rulings ​discount extensive evidence of user harm," he said.

The EU has long been among the most aggressive regulators in reining in tech giants. However, the strong regulatory environment they've created comes with a catch: Compared to the US and China, the region largely lags behind in homegrown AI innovation, adoption and investment. Though some US states could attempt similar legislation, such as California with the Transparency in Frontier Artificial Intelligence Act, the US federal government is unlikely to follow Europe's lead. Nevertheless, EU regulations can often have downstream effects because tech companies don't like playing by two sets of rules. As a result, the adjustments they make to comply with European regulations can affect products globally.  And in the best cases, such as universal USB-C adoption, it can have a positive impact for users. Forcing Android to give users a choice of their favorite AI, rather than Google Gemini always having preferred placement, could yield a similar result. Next, we have to wonder whether the EU will target Apple for similar AI dynamics now that Siri is about to become more capable and even more embedded on the iPhone. 

TOGETHER WITH CHECKSUM

Don't stop at CI/CD. Ship faster with continuous verification.

63% of engineering teams now ship code faster with AI. 72% have already had a production incident from AI-generated code. The bottleneck didn't disappear. It moved downstream.

Checksum is an AI-native continuous testing platform that auto-generates and self-heals your E2E test suite, runs inside your existing CI/CD pipeline, and keeps pace with the velocity you're already getting from AI coding agents.

Clearpoint Strategy saves $500k a year. Postilize ships 30% faster.

Setup takes under 30 minutes.

PRODUCTS

Anthropic and 1Password close agent password gap

AI agents are only as useful as the access you’re willing to give them. 1Password wants to make that access safer to grant.

On Thursday, the security company launched 1Password for Claude, a browser integration that gives Claude access to stored credentials without actually exposing the credentials to the model, memory, or Anthropic. The aim is to give Claude agents the access they need without exposing credentials or requiring the human to manually authenticate at every step, Nancy Wang, CTO at 1Password, told The Deep View.

“Agents are already performing sensitive actions on behalf of people, and the question is no longer whether that happens. It's whether it happens with the right controls in place,” said Wang. “1Password for Claude was designed for exactly this moment. The agent gets what it needs to securely complete the task.”

1Password explains that the integration gives agents the permissions they need, specific to the task at hand and the session, ensuring they do not carry over to other sessions and that confidential information is encrypted until it is inserted into the page by 1Password. Other safeguards, as listed in the post, include:

• User biometric approval: Claude requests task credentials, which users approve or deny via a biometric prompt.

• Agentic Mode: With Agentic Mode, whenever an AI agent takes control of the browser, 1Password automatically locks down, making everything in the 1Password vault unreachable except the credentials granted to the specific task.

• Field analysis: 1Password continues to scan the page after every autofill to make sure no confidential information remains exposed.

Agentic Mode is also available to all 1Password users, highlighting the interoperability of these safety measures and the aim to make all AI interactions safer beyond Claude: “This infrastructure isn't scoped to Claude. It's built to extend to other browser-based agents as the ecosystem grows, because we're not building for the integrations that exist today. We're building for the agent landscape that's coming,” added Wang.

1Password for Claude is available to 1Password business, family, and individual users on Mac. To get started, users need to install the 1Password and Claude desktop apps and browser extensions

While it’s best practice to keep credentials private from agents, that’s becoming increasingly disadvantageous for working professionals. It’s through those credentials that AI tools gain access to dashboards and data, unlocking a new level of helpfulness. That’s why developing secure ways to do it is the utmost priority. Users have relied on 1Password and similar apps to share passwords among co-workers for years, and today, AI agents are essentially another digital co-worker. So, any way to share those credentials securely is extremely valuable. But it’s important to understand this is still different from handing an AI your credentials outright, and that will likely never be a good idea.

TOGETHER WITH KALSHI

Think You Can Pick The World Cup Winner?

Then it’s time to put your money where your mouth is with Kalshi. Their CFTC-regulated exchange lets you take a position on who will lift the trophy, as well as the spread, game day events ($10 on a rain delay, anyone?), and even player props – like Messi getting another goal or assist to continue his ridiculous form. 

Ready to get in on the fun this weekend? Sign up for Kalshi right here and get $10 when you trade $10.

GOVERNANCE

New US open model takes on China's top AI

China and Europe have typically led the charge on open models. Now, another US company is joining the fray with a significant entrant into the race.

On Wednesday, Thinking Labs, the AI startup founded by ex-OpenAI CTO Mira Murati, launched its first model: Inkling. The model sets itself apart from leading AI labs in the US in a very important way: it is open-weight, allowing customers to customize it directly to their needs through fine-tuning and training. 

To suit that purpose, the model was built to perform across a wide range of areas. Or, as Thinking Machines describes it, it's a "generalist model" that can reason about text, images, and audio rather than being proficient in just one domain. This can be observed in benchmark performance, which isn't best-in-class but is consistent across specialties, as seen in the image below. 

Other model features include:

• Parameters: Mixture-of-Experts transformer model with 975 billion total parameters, 41 billion active.

• Context window: Up to one million tokens 

• Pretrained: On 45 trillion tokens of text, images, audio and video, allowing it to be proficient across all three. 

• Lightweight counterpart: Inkling-Small, a lighter-weight model with 12 billion active parameters.

• Fine-tuning: To make customization accessible, the company says it is making Inkling available on Tinker, the company's training API platform. 

• Cost-efficiency: Inkling's "controllable thinking effort" allows users to customize the cost/performance curve.

Until now, most open-source models have been developed in other countries, with China leading the way. These models are good for the ecosystem because they allow people and organizations to more deeply specialize the models, adjust their behavior, run them locally, and even save money. Though, as Thomas Randall, Research Director at Info-Tech Research Group, told The Deep View, they may not be for everyone. 

"If organizations have use cases where cost control, sovereignty, fine-tuning, deployment flexibility, or domain-specific performance matter, then open-source models are relevant," said Randall. "For the highest-risk or most complex work (e.g., legal, medical, cyber, or regulated decisions), closed frontier model providers are preferable because of safety tooling, support models, auditability, and more mature deployment ecosystems." 

Still, for the US to compete in the open-weights model space, it has a lot left to do, despite recent entrants from not only Thinking Machines but Nvidia, Google and OpenAI. A closer look at benchmark performance shows that many of China's leading labs, which have also taken a generalist approach with their open-weight models, have outperformed Inkling on benchmarks. That includes GLM 5.2, DeepSeek V4 Pro, and Kimi K2.6. Kimi K3 also just released on Thursday, and it's Kimi’s most capable model to date, with 2.8 trillion parameters.

Competition is the lifeblood of any industry. This is especially true for the state of AI we are in now, where enterprises and AI builders are prioritizing cost, and open-weight models provide a tangible solution while applying pressure on the leading labs to be more cost-conscious. Just this month, OpenAI and Meta have both released new models, each highlighting their cost efficiency. Still, it is important to keep in mind the broader backdrop, including the rising costs of finite resources such as data centers and energy. Open-weight models still need the same compute power, even if they cost less than proprietary models.

LINKS

• Stanford-Harvard study evaluates medical AI tools doctors can trust

• University of Houston professor uses AI to make roads safer in study 

• Anthropic research reveals agentic misalignment in frontier models

• Agent startup Runta raises $20M seed at $100M valuation

• Suno allegedly scraped decades of music and podcasts for AI training

• Meta alerts parents when teens discuss suicide, self-harm with AI

• Google renames NotebookLM to Gemini Notebook, adds features

• Google Search: Users can connect services, such as Instacart and YouTube Music

• Grok Build: SpaceXAI's coding agent and TUI are being open-sourced

• Claude Code: Artifacts can now call MCP connectors

• Gemini Spark: Now rolling out to Google AI Ultra subscribers in more countries and languages

• OpenAI: AI Deployment Engineer

• Hello Patient: AI Agent Product Manager

• Silna: AI Deployment Analyst

• Starbridge: AI Engineer

GAMES

A QUICK POLL BEFORE YOU GO