AI Leadership Weekly

Issue #30

Brad
May 08, 2025

Welcome to the latest AI Leadership Weekly, a curated digest of AI news and developments for business leaders.

Top Stories

Source: OpenAI

OpenAI remaining “non-profit,” sort of

OpenAI has announced a major structural shift, transitioning its for-profit LLC to a Public Benefit Corporation (PBC) while maintaining nonprofit oversight. This move aims to secure the massive funding needed for artificial general intelligence (AGI) development, which OpenAI estimates could require hundreds of billions—and eventually trillions—of dollars. The nonprofit will remain in control of the PBC, ensuring its mission to democratise AI access remains central.

The decision followed discussions with civic leaders and state attorneys general in Delaware and California, who emphasised the need for accountability in AGI development. OpenAI CEO Sam Altman outlined three key goals in a letter to employees:

1) Making AI tools broadly available to humanity

2) Transforming the nonprofit into the “most effective in history” for leveraging AI’s societal impact

3) Prioritising safety and alignment in AGI to ensure it benefits all.

Altman stressed, “We believe this is the best path forward—AGI should enable all of humanity to benefit each other.”

The PBC structure aligns OpenAI with peers like Anthropic and Patagonia, which balance profit with public good. Unlike OpenAI’s previous “capped-profit” model, the new structure simplifies capital access without compromising mission. Altman noted the nonprofit will become a major shareholder in the PBC, using its growing resources to fund initiatives in healthcare, education, and more. He added, “Creating AGI is our brick in the path of human progress; we can’t wait to see what bricks you will add next.”

Source: Pexels.com

250 CEOs want Comp-Sci/AI taught as K-12 “core components”

Over 250 CEOs—including Microsoft’s Satya Nadella, Etsy’s Josh Silverman, and Uber’s Dara Khosrowshahi—signed an open letter in The New York Times demanding that AI and computer science become "mandatory for every student".

Their message? The U.S. is “falling behind” countries like China, Korea, and Singapore, which already require every student to learn these skills. The letter warns that without action, American kids will be “AI consumers, not creators.”

The CEOs argue that making AI/CS core curriculum is vital for “keeping America competitive.” They highlight that rivals like Brazil and Singapore have already moved the needle, while the U.S. lags in preparing students for an AI-driven future. The letter arrives as President Trump’s administration pushes an executive order to integrate AI into K-12, higher ed, and workplaces via industry-academia partnerships.

Source: Pexels.com

MTA using AI to spy on New York subway

The MTA is teaming up with tech companies to test AI software that scans subway camera footage for “problematic behavior” and alerts the NYPD in real time. MTA Chief Security Officer Michael Kemper called it “predictive prevention,” aiming to flag potential crimes—like someone acting “irrationally”—before they unfold. “AI is the future,” Kemper said during a recent meeting, as the agency grapples with subway violence, including unprovoked attacks.

The system will analyse behaviour, not people, the MTA insists. A spokesperson emphasised that facial recognition won’t be used, focusing instead on actions like erratic movements or suspicious activity. About 40% of platform cameras are already monitored live, and the MTA has rolled out cameras in every train car since 2021. But critics aren’t convinced.

Governor Kathy Hochul has pushed for more surveillance since taking office, and the MTA’s plan comes amid pressure from federal officials. Transportation Secretary Sean Duffy threatened to withhold funds unless the MTA addresses crime and homelessness. While major felonies have dipped to pre-pandemic levels, assaults remain up, and 10 subway murders were reported last year.

And civil rights groups are sounding alarms. The New York Civil Liberties Union’s Justin Harrison called AI surveillance “notoriously unreliable and biased,” warning it could deepen disparities and create a “surveillance state.” “Real public safety comes from investing in communities, not omnipresent cameras,” he said.

In Brief

Market Trends

OpenAI is acquiring Windsurf IDE

As reported by Bloomberg, OpenAI has agreed to buy Windsurf, an AI-assisted coding tool (formerly Codeium), for $3 billion.

This would mark OpenAI’s most significant acquisition yet, aiming to bolster ChatGPT’s coding capabilities amid intensifying competition in the AI coding space. The deal has not yet closed, with OpenAI and Windsurf declining to comment.

Windsurf’s valuation has surged from $1.25 billion in mid-2023 to $3 billion, driven by recent funding talks with investors like General Catalyst and Kleiner Perkins. OpenAI’s purchase aligns with its strategy to expand coding features, a key area where rivals like GitHub Copilot and Amazon’s CodeWhisperer are also active. The company has previously acquired startups such as Rockset to enhance infrastructure for enterprise tools.

The move comes as OpenAI’s user base grows rapidly, surpassing 400 million weekly active users in February, up from 300 million in December. Simultaneously, the company is preparing to raise $40 billion in a new funding round led by SoftBank, at a staggering $300 billion valuation.

Anthropic launches its “AI for Science Program”

Anthropic is awarding up to $20,000 worth of free API credits to researchers in biology and life sciences via its new "AI for Science Program, aiming to speed up everything from drug discovery to agricultural breakthroughs. The initiative—as discussed in their CEO Dario Amodei’s book "Machines of Loving Grace"—offers AI tools to scientists tackling global challenges like genetic analysis and climate-resilient crops.

The program targets projects where AI can “meaningfully accelerate” research, such as modelling complex biological systems, parsing genetic data, or designing experiments. Anthropic’s API will act as a digital lab partner, helping researchers analyse data, generate hypotheses, and write grant proposals. The company emphasises that this isn’t just about profit—it’s part of their mission to build AI that “brings value to humanity.”

Applications are open to researchers at accredited institutions, with selections based on potential impact and AI feasibility. From there, Anthropic’s team, including field experts, will review the proposals and make the final decisions.

AI is DDoS-ing security researchers

For well over a year now, there have been ongoing reports of Github maintainers being flooded with pull requests by new and inexperienced developers, only to have those pull requests turn out to be pointless or contain issues that are completely made up.

This had been traced back to several coding academies encouraging their students to pad their resumes by "contributing" to open-source projects. This then led many to employ AI coding tools to generate bug reports, which the humans then blindly submitted. The maintainers then burn time engaging with what they think is a legitimate pull request, but find themselves unable to replicate the problem because...it never existed in the first place.

The linked video from Low Level repeats the same story where the maintainer of CURL engages in a back-and-forth trying to understand the problem, and only realises that the bug report was AI generated when it describes an issue with a function that doesn't exist. That is, the AI made up a function, then claimed it caused a security vulnerability within CURL.

At its least problematic, this practise annoys a few repo maintainers and wastes several hours. But, as more people try to claim bug bounties and pad out their Github profiles, this floods legitimate maintainers—who are often solo developers or on very small teams—with an impossible level of work. It has also been used as a vector for bad actors to gain access to vital, widely-used projects and inject their own malware.

Tools and Resources

Windsurf Wave 8
See why OpenAI was so exciting to acquire this company by checking out Windsurf's latest update.

Parakeet
Try out NVIDIA's new open-source model that gives you high-quality transcriptions.

Clova
Generate rough cuts of your videos with prompts.