POLICY

Trump to challenge state AI laws

Federal AI regulation is on the way, and it spells trouble for state lawmakers. 

President Donald Trump said on Monday that he intends to sign a “one rule” executive order preempting state AI regulation, confirming reports from late last month. In a post on Truth Social announcing the executive order, Trump said that AI will be “destroyed in its infancy” if forced to comply with state AI regulations. 

“We are beating ALL COUNTRIES at this point in the race, but that won’t last long if we are going to have 50 States, many of them bad actors, involved in RULES and the APPROVAL PROCESS,” Trump said in his post. 

The leaked version of the executive order, which began circulating in late November, would give the Department of Commerce the authority to cut funding to states that seek to regulate AI, and task Attorney General Pam Bondi with spinning up an “AI Litigation Task Force” focused entirely on challenging state AI laws.

In a post on X, White House AI and Crypto Czar David Sacks denied that the order is an AI moratorium, but rather “an attempt to settle a question of jurisdiction,” pushing back against “the most onerous and excessive state regulation.” 

“At best, we’ll end up with 50 different AI models for 50 different states – a regulatory morass worse than Europe,” Sacks said in the post. “This will stymie innovation, especially by small startups who can’t afford the compliance burden. Meanwhile, China will race ahead.”

In recent months, states like California, Colorado and Texas have sought to rein in the risks posed by the ever-evolving tech, passing regulation that targets risky model behaviors such as bias and discrimination, as well as curbing children’s use of AI. Laws like these could be rendered entirely moot by the executive order, giving model providers the power to self-regulate and govern.

The AI legal landscape in the US is just one example of a broader push towards deregulation.  In late November, the European Commission unveiled a “digital omnibus” package to loosen its AI and privacy regulations, which includes sweeping changes to the General Data Protection Regulation, the region’s watershed personal privacy law.

Though the US and EU are stripping back regulation to different degrees, both of these efforts highlight the frenzied fear of falling behind in the AI race. The EU, which was once a paragon of tech regulation that put citizens first, is feeling the pressure to step up as even its most impressive models don’t compare to those from the US and China. 

With little regulatory red tape, model developers will only be held to the standards that they set for themselves. And while some efforts may be put in place to avoid a PR nightmare when something goes awry, these firms will likely continue to take the Silicon Valley ethos of “move fast, break things” to heart, prioritizing innovation at all costs over safety.

TOGETHER WITH MONDAY.COM

The best AI for work

Meet monday.com, the AI-powered work platform built for the way modern teams really work / built for the way modern teams really work.

With AI woven into every workflow, monday.com keeps work moving — helping teams stay aligned and deliver faster, while giving leaders the clarity and control they need to drive real results.

Loved by teams. Trusted by leaders. See why monday.com is the best AI for work.

MARKETS

What bubble? AI valuations continue to soar

The two-month-old startup Unconventional AI raised $475 million at a $4.5 billion valuation in a round led by a16z and Lightspeed Venture Partners.

Unconventional AI, which was founded by Databricks’ former head of AI and aims to build more efficient chips, is the latest AI startup to garner an eyewatering valuation on the back of seemingly endless demand for compute. And while the ever-richer funding deals have drawn plenty of warnings of an AI bubble, some analysts think the blockbuster valuations are justified.

Even so, Unconventional’s seed round numbers are certainly unconventional. Carta's head of insights, Peter Walker, said the startup’s $475 million in seed funding is 121 times the median raise amount for seed rounds this year.

“Don't compare VC for chip companies and research labs to anything else if you want to stay sane,” Walker said.

These are indeed lucrative times to be in the chip or AI research lab business. NVIDIA continues to break records. Thinking Machines, the lab founded by former OpenAI executive Mira Murati, reached a $12 billion valuation in its seed round, even though it was still pre-product. And then, of course, there are the frontier labs OpenAI and Anthropic, last valued in private rounds at $500 billion and a reported $350 billion, respectively.

Taken together with some of the circular deals being cut by AI’s heavyweights, some point to frothy valuations in the sector as evidence of a speculative bubble that won’t be sustainable in the long run. Hedge fund icon Ray Dalio just told CNBC he believes we are in a bubble, and that the biggest benefits of AI would go to users who cut costs with AI rather than to hyperscalers spending to build AI.

Others point out that although AI valuations are unprecedented, the math still sort of works out.

“The only bubble in [AI] is in bubble talk, in my opinion,” Morgan Stanley head of applied equity advisors Andrew Slimmon said in a recent memo. The dot-com bubble popped once the demand for internet services and products no longer outstripped the industry’s ability to supply them, he noted.

“In listening to Q3 quarterly commentary from [AI-related] companies, what struck me was their inability to keep up with demand,” Slimmon said.

Multi-billion dollar seed rounds are like a Rorschach test for the AI industry. They’re either a result of the transformative power of AI or evidence that the market is ripe for a major correction. Both things can be true, however, and it would be foolhardy to bank on imminent AI collapse. As Dalio noted on CNBC, just because you’re in a bubble doesn’t mean the bubble is about to pop. But ultimately, megarounds like Unconventional’s may have a more banal explanation — there’s a lot of venture capital sitting around right now, and investors don’t have enough outlets to deploy it.

TOGETHER WITH FLORA

This Platform Has Every Creative AI Tool, All In One Place

It’s called FLORA, and it isn’t your standard creative platform. FLORA is like the Photoshop to every other platform’s MS Paint – sure, they’re both technically for design, but one offers a laundry list of best-in-class tools trusted by pros everywhere, while the others… well, the others can be kinda fun to goof around in.

Designed for professionals who demand control, speed, and precision (all without sacrificing creative control), FLORA’s seamless workflow enables iteration and collaboration at scale. It’s the first AI tool built by creatives, for creatives – and now, you have a chance to try it. The first 50 people to use code DEEPVIEW at checkout will get 50% off their first month of FLORA's Agency Standard Tier (monthly option only), so get moving right here.

CONSUMER

AI giants grapple with the ad question

AI firms search high and low for ways to monetize their models. Though digital ads have typically been a moneymaker for tech giants, integrating them might not be so easy.

AdWeek reported on Monday that Google told advertisers that it plans to bring ads to its Gemini chatbot. Though few details were revealed, Google representatives told ad clients the company is targeting a 2026 rollout of ad placements in Gemini, a plan separate from ads in AI Mode in Search, according to the report.

Google’s VP of Global Ads, Dan Taylor, denied the report, claiming in a post on X that “there are no ads in the Gemini app and there are no current plans to change that.” Google did not offer additional comment to The Deep View.

The news comes as OpenAI faced scrutiny over rumors that the company is implementing ads into ChatGPT, following screenshots from users circulating on X depicting promotional materials for companies like Target and Peloton in responses to their chats. 

• OpenAI disabled promotional messages in chat following the backlash, with Chief Research Officer Mark Chen saying, “anything that feels like an ad needs to be handled with care, and we fell short.” 

• Nick Turley, OpenAI VP and head of ChatGPT, said in an X post that there are no live tests for ads in the chatbot.

Though both of these AI giants have vehemently denied plans and tests to bring ads to their large language models, leaning into advertising and shopping may be a way to help these firms recoup some of the billions that they’ve spent on their models. Social media firms like Meta and TikTok have raked in billions from digital ad revenue, and Google’s search legacy sits atop a digital ad treasure trove. Given that OpenAI, in particular, has upwards of a trillion dollars committed to the historic data center buildout, any path to cash may be worth exploring

However, if the public’s reaction to this has told us anything, it’s that bringing digital ads to chatbots may cause some friction with users. While these firms could risk diminished user numbers, this friction itself could be a boon: Just like many chose to spend on premium tiers of streaming services, bringing in ads could lead some to opt for premium tiers of AI models if ads are a part of their futures.

LINKS

• Robotics startup Skild in talks to raise more than $1 billion at $14 billion valuation

• Palantir CEO Alex Karp launches Neurodivergent Fellowship following interview

• Young workers in Kenya are training AI models for major Chinese tech firms

• Google, Nextera Energy partner on data center development

• OpenAI, Instacart launch grocery shopping experience inside ChatGPT

• IBM to acquire data infrastructure company Confluent for $11 billion

• Workers are saving up to an hour a day using AI, according to OpenAI survey

• Nvidia wins right to export H200 chips to China

• Essential AI Rnj-1: An open source model that’s on par with those from Google and Alibaba. 

• Z.ai GLM-4.6V: The newest vision-language model from Z.ai, available both in 106 billion parameters and 9 billion parameters for lightweight, local workloads. 

• Mixboard Updates: Google’s AI-powered concept board now includes access to Nano Banana Pro, new file types and multi-board projects. 

• Claude Code for Slack: Anthropic has launched its popular AI coding tool inside of workplace messaging platform Slack.

• OpenAI: Residency 2026

• Meta: AI Specialist - Product and Applied Research

• Google DeepMind: Research Scientist/Research Engineer, AI for Secure Code

• xAI: Member of Technical Staff, Multimodal Understanding

GAMES

A QUICK POLL BEFORE YOU GO