🪙 OpenAI Pledges $250M for Shared AI Prosperity

OpenAI just pledged $250M to help people whose jobs its own software is making redundant. Read that twice, because the company is now funding both sides of the same problem. Let's dive in!

In today’s insights:

LATEST DEVELOPMENTS

Evolving AI: OpenAI commits $250M to steady the economy as AI reshapes work worldwide.


Key Points:

The OpenAI Foundation committed an initial $250M to what it calls secure and abundant economic futures. The money first builds tools to track how AI shifts wages and jobs, then funds help for displaced workers. From there it gets ambitious, testing ideas once treated as fringe: shifting taxes from labor to capital, windfall mechanisms, and sovereign wealth funds modeled on Norway and Alaska that hand citizens a direct stake in AI-driven growth.

The company writing the relief check is the same one writing the software behind a real share of this year's layoffs. That is not a knock, it is leverage. By funding the measurement and the policy menu at once, OpenAI gets to shape which fixes get tested before the disruption it is accelerating fully arrives.

Evolving AI: Adobe just launched Firefly AI Assistant inside Adobe Firefly, its all-in-one creative AI studio.

Powered by Adobe’s new creative agent, the assistant orchestrates creative workflows across apps like Photoshop, Premiere, Lightroom, Firefly and more.

Instead of switching between apps or figuring out which tools to use in what order, you simply describe what you want to create while the assistant coordinates the workflow behind the scenes.

Creative work is quickly shifting from “how do I make this?” to “here’s what I want.” Adobe’s Firefly AI Assistant is one of the clearest examples of that shift yet.

Explore Firefly AI Assistant →

Evolving AI: DeepMind's CEO just pulled his AGI estimate forward, now floating 2029 as a real possibility.

Hassabis feels more confident the industry is on the right path, so his range has narrowed. He still expects AGI around 2030 but now sees 2029 as possible, and says we're standing in the "foothills of the singularity." The coming wave of AI agents, he argues, is a societal stress test for the bigger systems still to come. He flagged recursive self-improvement, where systems start speeding up their own development, as a milestone every lab is watching. He also backed a possible executive order to test new models before release.

Why It Matters:

For a while the short timelines came from Amodei and Musk while Hassabis held the careful line. So when even the cautious voice pulls his number forward, the field's center of gravity moves with him. The telling part is that his urgency is aimed at governments and economists rather than rivals, which suggests the bottleneck he's worried about isn't the technology catching up to us, but us catching up to it.

Lindy reads your email as it arrives, drafts replies in your voice, and texts you a meeting brief before every call. Works over iMessage. One minute setup. Try it free.

Try Lindy for free

Evolving AI: A critical Starlette flaw puts millions of AI agents at risk worldwide.

Researchers found a critical flaw in Starlette, an open source framework downloaded 325 million times a week. Because Starlette is the base of FastAPI, the bug spread far and wide. Nicknamed BadHost, it lets an attacker slip a single character into an HTTP header and skip authentication entirely. From there, they can reach servers, MCP connectors, stored credentials and sensitive data. A patch landed in version 1.0.1, though plenty of systems are still running old code.

MCP servers store credentials for the systems agents touch, from inboxes to identity databases, which makes them a high-value target. They logged near zero flaws before 2025 and roughly 95 last year. So a single weak link in foundational code now puts a huge slice of the AI stack within reach at once. The agent boom keeps widening the attack surface faster than teams can patch it.

QUICK HITS

💰 Cognition raises $1B at $26B, more than doubling in eight months.

🏭 Nvidia bets $150B a year on Taiwan as its AI manufacturing hub.

🛡️ BNP Paribas leans on Mistral as Anthropic's Mythos rattles banks.

💸 The VC who backed Anthropic at zero revenue turned $75M into $3B.

 *partner link