POLICY

AI’s future may hinge on who writes the rules

Some of the most prominent voices in AI have strong ideas about the way the tech should be governed. 

At the G7 Summit in Évian-les-Bains, France, on Wednesday, Anthropic CEO Dario Amodei and Google DeepMind founder Demis Hassabis called for a US-led coalition to shape standards for AI, according to CNBC and The Financial Times. 

The tech leaders proposed that this coalition be an international effort with the US leading the conversations around AI's risks. They warned of the tech's dangers, including major bioterrorism and cybersecurity risks, if an alliance couldn't be formed to preempt them.

• The proposal was supported by OpenAI's Sam Altman and Canadian Prime Minister Mark Carney, according to media reports. 

• According to the FT, Amodei told G7 leaders they must “resist the temptation to splinter” over the deployment of advanced AI tools. 

The proposal comes days after Anthropic's latest messy dispute with the Trump Administration over the release of its latest models, Mythos 5 and Fable 5. The US government imposed export controls on the models, restricting their use outside the country and by foreign nationals, citing cybersecurity and national security concerns. 

As a result, other government leaders have expressed their concerns: According to the FT, French President Emmanuel Macron said that if the US can "turn off the switch" on powerful models, the impacts could ricochet back on the AI companies leading the market. Indian Prime Minister Narendra Modi also expressed his concern, reportedly saying that democratic nations must have access to leading AI models to safeguard infrastructure.

Macron also warned against the risks of “non-cooperation among democracies" as it relates to AI regulation, calling for stronger regulation to "jointly define common standards."

If major model providers are to be believed, AI will be a fundamentally transformative technology for society. I've written that the power of this technology has concentrated in the hands of an elite few companies, which has created risks regarding who exerts influence over this societal transformation. It's something that the leaders of these companies themselves have even started to recognize, with Altman, Amodei and OpenAI's Greg Brockman all pointing towards the risks of power centralization. Still, despite these leaders calling for international cooperation to temper the risks, the question of how this tech should be regulated is far from settled. If this technology turns out to be as transformative as these companies claim, it has the potential to have a broad impact on the global community. And that raises questions. Should the regulation of something so broadly transformational be placed in the hands of a small group of government and business leaders, or be voted on more democratically by the people it affects? And with technology changing and evolving so rapidly, how will any kind of regulation keep up? 

A QUICK POLL

TOGETHER WITH COLUMBIA BUSINESS SCHOOL EXEC ED

Finance professionals are using AI. Most don't know what they're actually doing with it.

The 8-week AI for Business & Finance Certificate Program from Columbia Business School Exec Ed + Wall Street Prep is built for non-technical roles and helps you cut through buzzwords, apply AI to your work, and lead adoption in your org.

• Apply AI directly to financial analysis, reporting, and decision-making workflows

• Collaborate in LIVE weekly office hours with Columbia faculty

• Earn a Columbia certificate and join a 5,000+ network of finance and business professionals

Use code DEEPVIEW to save $300 on tuition. Enroll today.

GOVERNANCE

Why open ecosystems may beat AI agent lock-in

Nearly all of the big tech and AI companies are racing to solve the problem of AI agents in the enterprise. 

The Deep View has covered the latest attempts from Microsoft, Snowflake, AWS, Anthropic, OpenAI, Perplexity, and others. 

The reason they are all racing to play a role is that AI agents are creating a multitude of problems:

• Data leakage

• Compliance issues

• Privacy challenges

• Security vulnerabilities

• Inaccuracies 

• Cost overruns 

But most of all, agents aren't living up to the promise of automating workloads and creating enough value to justify the massive token costs that agents are running up in most businesses.

And now, the fact that all the vendors are offering their own agent solutions is starting to cause a new set of problems due to lock-in and a lack of cross-platform compatibility.

"The SaaS providers are now each providing you an agent that they want you to use," said Databricks CEO Ali Ghodsi on Tuesday at the keynote of Databricks Data + AI Summit. "They have agent builders that they want you to use. They have their own MCP servers… So it's kind of a quagmire. It's a complete mess." 

Databricks' answer to the problem is, naturally, its own agent, called Genie. But the real answer is the principle that Databricks always seem to circle back to: open ecosystems. 

In an interview with The Deep View, Databricks chief AI scientist Jonathan Frankle said, "The Databricks team has worked really hard over the past decade to make sure things are open. The way I frame it to customers is that I have to win your business every day, because if I don't win your business every day, you can leave… You can just take all this data straight out of Databricks. We don't really have much lock in, and that's kind of the point."

The full Genie ecosystem was announced at Data + AI Summit and includes Genie One for querying data and figuring out tasks, Genie Ontology to give your AI the business context to solve big problems, Genie App Builder for vibe coding business solutions, Genie Agent for reusable, templated agents, and Genie Code for more advanced engineering workflows.

I have to hand it to Databricks for creating solutions that are very polished and easy-to-use while always leaning heavily on an open ecosystem foundation. But the enthusiasm of the community building around Databricks at the Data + AI Summit was what impressed me the most. The customers and partners I talked to loved that they aren't trapped or locked into the Databricks ecosystem. They also talked about Databricks as a partner and collaborator more than a technology vendor that has all the answers to their problems. That's not easy considering a lot of Databricks customers are businesses that have been around for decades and have a ton of legacy systems to manage.

TOGETHER WITH SHOPIFY

Five apps Shopify built within days

As part of the Spring ‘26 Edition, a small group of Shopify designers and developers built five shopping apps in a few days that make everyday activities shoppable, with commerce woven into the experience itself.

One takes a trip itinerary and creates a shopping list. Another suggests products based on your horoscope. All run on Catalog API to query billions of merchant products and Universal Commerce Protocol for the full commerce journey.

They show what's possible for developers, designers, and anyone with an idea.

See what else shipped.

CONSUMER

Evan Spiegel's bet on life after the smartphone

Snap CEO Evan Spiegel joins The Deep View Conversations to discuss SPECS, Snap's long-awaited augmented reality glasses and why he believes they represent a new era of computing.

Senior reporter Sabrina Ortiz interviewed Evan at Augmented World Expo immediately after the SPECS unveiling and Spiegel explained why Snap spent more than a decade building toward this moment, how SPECS differ from AI smart glasses and mixed reality headsets, and why he sees AR glasses as the future beyond smartphones.

Other topics covered include:

• Why Snap calls SPECS a "computer" instead of AI glasses

• How SPECS combine wearability with advanced spatial computing

• The role AI played in making consumer AR glasses viable

• Why shared experiences could become AR's killer app

• The challenge of competing with Apple, Meta, and other tech giants

• Snap's 12-year investment in augmented reality hardware and software

• The importance of developers in building the AR ecosystem

• Why Spiegel believes people are ready for an alternative to smartphones

• How AR glasses could make computing more human

Spiegel argues that after nearly two decades of smartphone dominance, consumers are increasingly looking for a more natural way to interact with technology. Snap's bet is that augmented reality glasses can bring computing into the world around us instead of pulling us away from it.

📺 Watch on YouTube.

🎧 Listen in your favorite podcast player.

Subscribe to Deep View Conversations for interviews with the leaders shaping the future of AI, business, and technology: tdv.transistor.fm

LINKS

• Shoe company Allbirds rebrands as Smartbird amid AI pivot

• AI medical tool Mira, Google's Amie surpassed doctors at diagnostics

• Estonia will grant digital ID numbers to AI agents to control access

• Uber to expand premium robotaxi service to Houston next year

• AI military startup Twenty raised $100 million at $1 billion valuation

• World models startup Odyssey raises $310 million Series B

• Vercel Eve: An open-source agent framework for building, running, and scaling agents.

• Unreal Engine 5.8: The 3D design platform now has experimental MCP server support. 

• Claude Design: Anthropic's AI design tool has gotten an overhaul, including a fix to how fast it burns through tokens. 

• ChatGPT: OpenAI's flagship chatbot now has a faster, more reliable, and easier to manage scheduled tasks.

• Meta: Tech Lead, AI Research Scientist (Robotics)

• Anthropic: Research Engineer, Knowledge Team

• Tencent: Hunyuan Multimodal Algorithm Researcher (Omni-Modal)​​

• Google: Image Processing Engineer

GAMES