Anthropic, OpenAI ‘Pacing’ Spooks Some Startup Customers

It’s not clear what impact the pledges by Anthropic and OpenAI to slow the pace of AI development will have on their upcoming public offerings. But the comments are having real effects already: more people outside Silicon Valley and Washington are now alert to potential risks from AI in a way they weren’t even very recently. That’s not great news for startups trying to convince consumers and businesses that the tech is worth it, some investors told me.

One venture capitalist, for example, told me about a startup that has recently found customers to be more wary of buying its AI app for small- and medium-sized businesses in the light of news stories about potentially catastrophic AI.

That investor also thinks that potential government regulation or self-monitoring may favor the biggest startups, which have the financial resources, tools and employees to make sure their AI isn’t about to wreak havoc. Already the biggest labs are discussing collaboration to create a standards body for the AI industry, as my colleague Leo reported over the weekend. Representatives from Anthropic, OpenAI and Google have been meeting regularly since July about potentially forming an AI safety standards body, OpenAI’s global policy chief Chris Lehane confirmed Tuesday.

That could make it harder for neolabs, or the startups that attempt to research new approaches to developing AI models, to compete, said the investor.

On the other hand, one investor in early-stage software companies thinks the scrutiny on the frontier labs will be a boon to startups building tech that uses or supports the models. Expect VCs to flock to tech that improves model security or infrastructure, he added.

We’re already reported on several such funding rounds. These include Gimlet Labs, which routes AI work across the right kind of chips for the task at hand, and which raised three tranches of funding recently. Security startup Neo, which aims to protect devices from threats potentially posed by AI agents, raised $100 million from Andreessen Horowitz, Bessemer Venture Partners and others in July.

Meanwhile, another investor who has backed several applied AI startups said that frontier model pacing could buy app-layer startups time to make their tech better. If the labs are slowing down their rate of research and development, the investor said, perhaps an app company focused on a specific task or project can take advantage of the lull to make swift advancements. That could be a plus for apps like AI agent startup Instinct, which is looking to raise $1 billion in new funding as it competes with deep-pocketed rivals like Meta.

Today on TITV

Check out today's episode of TITV in which we speak with the CEO of Handshake about his perspective on moderating AI development.